- IAPC-OBP

Chapter
PHYSICOCHEMICALPROFILING
INDRUGRESEARCHANDDEVELOPMENT
KrisztinaTakács‐Novák
Contents
1.1. INTRODUCTION...........................................................................................................................................3 1.2. THEORETICALBACKGROUND..............................................................................................................6 1.2.1. Thephysical‐chemistryofdrugaction................................................................................6 1.2.2. Physicochemicalparameters...................................................................................................8 1.2.2.1. Ionization(pKa).............................................................................................................8 1.2.2.2. Solubility(logS)........................................................................................................17 1.2.2.3. Lipophilicity(logP)..................................................................................................20 1.3. METHODSFORPHYSICOCHEMICALPROFILING.......................................................................25 1.3.1. pKadetermination......................................................................................................................25 1.3.1.1. Potentiometricmethod..........................................................................................26 1.3.1.2. UV/pHtitration..........................................................................................................27 1.3.1.3. Othermethods............................................................................................................28 1.3.1.4. Co‐solventmethod....................................................................................................30 1.3.1.5. Decisiontreeformethodselection...................................................................31 1
Chapter1
1.3.2. logSdetermination....................................................................................................................32 1.3.2.1. Methodsfordeterminationofkineticsolubility..........................................32 1.3.2.2. Methodsfordeterminationofequilibriumsolubility...............................33 1.3.2.2.1. Saturationshake‐flaskmethod(SSF).............................................33 1.3.2.2.2. Potentiometricmethods........................................................................34 1.3.2.2.3. μDISSmethod.............................................................................................35 1.3.2.2.4. Highthroughputmethods....................................................................35 1.3.2.3. Specialapplications..................................................................................................36 1.3.3. logPdetermination....................................................................................................................37 1.3.3.1. Shake‐flask(SF)method........................................................................................37 1.3.3.2. Potentiometricmethod...........................................................................................38 1.3.3.3. IndirectlogPmeasurementmethods..............................................................39 1.3.3.4. Highthroughputmethods.....................................................................................39 1.3.3.5. Decisiontreeformethodselection....................................................................40 1.4. CASESTUDIES............................................................................................................................................41 1.4.1. pKadetermination.......................................................................................................................41 1.4.2. logSdetermination....................................................................................................................44 1.4.3. logPdetermination....................................................................................................................49 1.5. OUTLOOK.....................................................................................................................................................52 Acknowledgement..............................................................................................................................................52 REFERENCES.........................................................................................................................................................52 2
1.1.
INTRODUCTION
Thepurposeofdrugresearchistodevelopeffective,safe,andhighqualitynew
medicines to treat diseases where no drugs or otherwise nonoptimal ones are
available.Thisactivityisverycomplex,lengthy,expensive,andrisky.Sincedrug
research became industrialized, the highest level of scientific and technological
knowledge has been applied during the given era. Fundamentally, the industry
usesandputsintothepracticethenewestscientificresultsasearlyaspossible
thusdrugresearchitself becomesthedrivingforceforthedevelopment ofnew
theories,technologies,andmethods[1].
Takingalookbackatthehistoryofdrugresearch,onecanrecognizeonthelong
way of the evolution of the present system some milestones, paradigm‐changes
whichresultedinconsiderabledevelopmentinitsage(Figure1.1).Inthe‘60sof
thelastcentury,theformerlyusedtraditionalmethods(suchastheextractionof
activecompoundsfrommedicinalplants;randomscreening,trial‐errormethod;
side‐effect observation; serendipity, etc.) more or less have been replaced or at
least extended by the new strategy of rational drug design. Its first application
wastheQuantitativeStructure‐ActivityRelationships(QSAR)analysisintroduced
by C. Hansch [2] and based on the accumulated knowledge of structure‐activity
relationships.Therationaldrugdesignwascompletedwiththeapplicationof3D
molecular modeling, theoretical and computational chemistry (Computer Aided
DrugDesign,CADD)andprovedtobeamoreeffectivetoolthanpreviousonesin
thediscoveryandoptimizationofnewactivemolecules.Theappearanceandfast
expansionofhighthroughputscreening(HTS)andcombinatorialchemistryinthe
‘90shavegreatlyenhancedthenumberofactivecompoundsfound[3].Thelatest
paradigm‐changewasprovokedbythehumangenomeprojectandtheincreased
number of potential targets identified by genomics. However, these changes in
the research strategy did not mean that former methods were completely
neglected,indeedamajorityofthemareaninherentpartofdrugresearch.Each
methodhasitsappropriateuseandimportanceinit.
Sincethefirstrecognitionsofstructure‐activityrelationships,medicinalchemists
involved in drug research have been always paying outstanding attention to
those properties of drugs which determine their pharmacological action. The
knowledge of solubility, ionization ability and lipophilicity of drug candidates
provides useful information about the expectable pharmacokinetic properties
3
Chapter1
and gives synthetic chemists adequate tools to improve them by modifying the
structuralmoietiesofthemolecule[4].
traditional
methods
rational drug design
QSAR
1960
CADD
1980
HTS
Combi
Chem
1990
genomics
proteomics
metabonomics
2000
Figure1.1.Strategiesindrugresearch
2010
In the past, however, the main focus of drug research was first devoted almost
exclusively to the pharmacodynamic aspects of the biological activity and only
laterinthe developmentphasewerethepharmacokineticpropertiesexamined.
Thishasledtoahighattritionrateofcompounds.Inthelate‘80sstudiesreport‐
edtwoprominentreasonsofdrugcandidatefailure:thepoorbiopharmaceutical
properties (e.g. low bioavailability) and safety. Pharmaceutical companies have
made initiatives to shift the physicochemical profiling of compounds earlier in
thedrugdiscoveryprocess[5].
Currently, drug research is usually divided into two main phases: (1) discovery
phase, which involves the target identification, hit discovery, lead selection and
optimization;and(2)developmentphaseinwhichpreclinicalandclinicalstudies
are conducted (Figure 1.2). The role and timing of the physicochemical
characterization has considerably changed. The new strategy applied since the
‘90s is based on a parallel optimization of efficacy and prognostic profiling of
drugability.Thisrequiredanewmentality:tobreakdownthewallbetweenthe
discovery and development phases and to migrate from sequentially assessing
efficacy and drugability to the parallel process; to evaluate the therapeutic and
drug‐likefeaturestogether[6].
Figure1.2.Drugresearchprocess
Goodpharmaceuticalproperties,besidestheefficacy,meangoodabsorptionand
distribution, chemical and metabolic stability (appropriate bioavailability) and
lowtoxicity.
4
Physicochemicalprofilingindrugresearchanddevelopment
Foroptimizationofdrug‐likeproperties,physicochemicalparametersaresimple
andcheaptoolsintheearlyphaseofdrugresearch.Determinationofproperties
relevanttobiologicalactivityofdrugssuchasionization,solubility,lipophilicity
andpermeabilityiscalledphysicochemicalprofiling(Kernsetal.2001.)[7].
It is distinguished from the more complex term pharmaceutical profiling which
involves the investigation of integrity, stability, metabolic properties (e.g. CYP
450 inhibition), transporter effects and drug‐drug interactions as well (Figure
1.3)[8].
pharmaceutical profiling
ionization
drug-drug
interactions
integrity
solubility
lipophilicity
permeability
metabolism
physicochemical profiling
stability
transporter
effects
Figure1.3.Pharmaceuticalprofilingvs.physicochemicalprofiling
The present chapter focuses on only three parameters of physicochemical
profiling (pKa, log S, and log P) while Chapter 3 is dedicated to the role and
determination of membrane permeability. The traditional non‐automated, time‐
andmaterial‐consumingmethodsdevelopedinthepastforphysicochemicalpro‐
filingarenotsuitableindiscoveryforthemeasurementofthedrasticallyincre‐
asednumberofnewchemicalentities(NCE).Nowadays,suchearlyphysicochem‐
icaldeterminationsmustbematerial‐saving,HT,andreasonablyreliable.Several
excellent commercial instruments have been developed for this purpose, which
areminiaturized,automated,andadaptedtohigh‐throughputtechnologies[9].
Thefirstcomprehensiveoverviewofphysicochemicalprofilingwasreportedby
P.Taylorin1990[10].Theprogressivedevelopmentachievedinthenextdecade
is surveyed in A. Avdeef’s book: Absorption and Drug Development: Solubility,
Permeability and Charge State [11]. This book can be considered as the most
competentanddetailedcompilationofadvancedknowledgerequiredbyphysical
chemists involved in drug development. Numerous reviews summarized the
state‐of‐the‐artofnewHTexperimentaltechniques[12‐15],themostrecentwas
5
Chapter1
publishedbyY.Henchozetal.[16].So,variousliteraturesourcesareavailablefor
all who would expand their understanding of physicochemical profiling
accordingtotheirneed.
Theaimofthischapteristoprovide:(i)aconcisesummaryoftheoreticalback‐
ground; (ii) a comparison of different experimental methods and approaches;
(iii) an introduction to ample, useful, and practical examples. The case studies
taken from more than 30 years of experiences of the author are intended to
providehelptophysicalchemistsintherightmethodselectionandmeasurement
ofdifficultmolecules.
1.2.
THEORETICALBACKGROUND
1.2.1. Thephysicalchemistryofdrugaction
Drug action is a consequence of several chemical and biological processes in
whichbindingtothereceptor(pharmacodynamicphase)isessential.Besidesthis
however, the pharmacokinetic processes have also fundamental importance in
the biological activity. The active ingredient of a drug must separate from the
appliedpharmaceuticaldosageform,mustdissolveinbodyfluidsandpermeate
throughbiologicalmembranestoreachthereceptorsite.Followingthereceptor
response, the active compound dissociating from the binding site generally
undergoes metabolism and is excreted from the body. These liberation, absorp‐
tion, distribution, metabolism, excretion (LADME) features are mainly deter‐
mined by the physicochemical properties of drugs, namely by ionization, solu‐
bilityandlipophilicity.
Biological membranes are the main physiological permeation barriers to be
crossedbydrugs.Structurally,theyhavealipidbilayerresultingfromtheorien‐
tation of amphiprotic lipids (phospholipids, glycolipids, sphyngomyelin) and
cholesterolintheaqueousmedium.Thisbilayerhassomeofthepropertiesofa
two‐dimensionalfluid(fluid‐mosaicmembrane model)inwhichindividuallipid
molecules can diffuse rapidly in the plane of their monolayer (lateral mobility)
butcannoteasilypasstotheothermonolayer.Animportantobservationisthat
phospholipids are asymmetrically distributed in the membrane. Generally, the
outer (extracellular) half of the bilayer comprises mainly zwitterionic lipids
(phosphatidylcholine andphosphatidylethanolamine),whereastheinner(intra‐
cellular) part contains negatively charged lipids (e.g., phosphatidylserine). Dif‐
ferent proteins that induce transporter, signal transduction, or metabolic func‐
tionsareintegratedintothelipidbilayer[17,18].Thebiologicalmembranesare
apolarbarriers,wheretherelativepermittivityinsideisextremelylow(ε~2).It
has long been assumed that most drugs use transcellular transport and pass
these barriers by passive diffusion which is favorable only for unionized,
lipophiliccompounds.Thereareseveralothermechanismsofpermeation.Active
transport is ligand‐mediated by different transporters for compounds. Paracel‐
lular permeation exists between the cells for smaller, more polar compounds.
6
Physicochemicalprofilingindrugresearchanddevelopment
Some compounds are transported by endocytosis, when the molecules are
engulfed by the membrane and move through the cell in these membrane‐
enclosed vesicles. For further detailed information, the reader is encouraged to
reviewspecializedresources[18,19].
ThepH‐partitionhypothesis[20]providesagoodmodelforthepassivetransport
ofionizablemoleculeswithsufficientlipophilicity.Figure1.4showsaschematic
representation of the transport of a basic (B) (e.g. papaverine, chlorpromazine,
etc.)andanacidic(HA)(e.g.acetylsalicylicacid,ibuprofen,etc.)molecule.Inthe
extracellularaqueousmedium,theratioofionized([BH+]or[A‐])andunionized
([B]or[HA])formsisdependentontheactualpHofthegivencompartmentand
the pKa of the compound. The uncharged, neutral species has much higher
lipophilicitythanitscharged(ionic)form,thusitcanpermeatethroughthelipid
membraneevenifbeing presentas aminorcomponent.Inmedicinalchemistry
thisspeciesiscalledthe“transportform”.Enteringintotheintracellularaqueous
phase,anotherionizationprocesstakesplaceresultingintheionizedformagain
whichgenerallyinteractswiththetargetandisreferredtoasthe“receptorform”.
Theamountofthetransportformpresentatthemembranesurfacedependson
itssolubility.Moleculesmustbeinsolutioninordertodiffuseintothemembra‐
nes,howeverlowsolubilitycanbealimitingfactorofpermeation.Permeability
asadeterminantkineticparameteroftransportisdiscussedinChapter3. AH
pKa
-
A +H
+
+
B+H
+
pKa
extracellular
BH
intracellular
AH
-
+
+
A +H
B+H
+
BH
receptor
Figure1.4.Transportandreceptorformsofanacidandabase
The concept derived from the pH‐partition theory that “only neutral molecules
permeatemembranes”startedtobequestionedfromthemid‘90sbecauseofan
increasing body of experimental evidence supporting ion‐partitioning into
artificialmembraneslikeliposomes[21].Thiswasinterpretedwithelectrostatic
interactions and hydrogen bonding between the charged group of compounds
7
Chapter1
and ionized polar head group of phospholipids in the “pH priston model” [22].
Recently, S. Krämer and coworkers [23] reviewed the mechanisms underlying
lipid bilayer permeation. They proposed a kinetic “flip‐flop model” based on a
three‐stepmechanism,namelythepartitioningintoonelipidlayer,translocation
(flip‐flop) to the opposite lipid layer and partitioning into the aqueous phase.
Accordingtothismodel,thepermeationofachargedspeciescouldbetheresult
of occasionally occurring trans‐membrane translocation of charged compounds.
It was concluded that membrane permeation is more complex than expected
fromasimplediffusionmodelandpH‐partitionhypothesis.
Anotherpossiblemechanismoftransportforionized,hydrophiliccompoundsis
carrier‐mediated active transport. The increasing number of different uptake
transportersdiscoveredinthepast15yearshighlightstheimportanceoftherole
of active transport in membrane permeation of drugs which may be under‐
estimated.Theirphysiologicalfunctionistodeliverthenecessarynutrientsand
other endogenous biochemical compounds having low lipophilicity for passive
diffusion to the cell. Several drugs were found to be the substrate of different
specific transporters like oligopeptide (PEPT1: captopril, enalapril, ampicillin,
acyclovir),organicanion(OATP1:fexofenadine,enalapril,temocaprilat),organic
cation (OCT1: metformin, famotidin), or nucleoside, etc. [18,24]. The efflux
transporters (P‐glycoprotein, P‐gp; breast cancer resistance protein, BCRP;
multidrugresistanceprotein,MRP2)assistinthemovementofcompoundsoutof
thecellastheyprotectthecell frompotentially toxicxenobiotics.Thisoutward
transporthasanegativeeffectonthepharmacokineticsofsomecompounds.The
activity of efflux transporters is very intensive in the blood‐brain barrier and
sometumorcellsresultinginmultidrugresistance.Bindingtothetransportersis
determined by the chemical structure of compound. Similar moieties to the
natural substrate, a large number of H‐bond acceptors (N + O atoms), and high
molecularweight(Mw>400)appeartoincreasethelikelihoodofP‐gpefflux[25].
Physicochemical properties influencing the fate of a drug in the body are
described by the thermodynamic equilibrium constants. Below, we summarize
thefundamentalsofpKa,logS,andlogPterms.
1.2.2. Physicochemicalparameters
1.2.2.1. Ionization(pKa)
Drugs are multifunctional compounds. A great majority of them contain one or
moreionizable(acidicorbasic)functionalgroups.Inaqueoussolutions,ionizable
compoundsexistindifferentionization(chargedoruncharged)statesdepending
ontheirstrengthofacidityorbasicityandthepHofthesolution.
Definitions,terms
Theionizationconstant(oraciddissociationconstant),Ka,isusedtocharacterize
the acid‐base chemistry of a molecule generally expressed as a negative loga‐
rithm:‐logKa=pKa.Inmedicinalchemistry,itiscommontousepKaforbothacids
8
Physicochemicalprofilingindrugresearchanddevelopment
andbases.Inaqueoussolutions,thepKascalespans from0to14. Thestronger
the acid, the lower is its pKa value. The opposite is true for bases; a higher pKa
valuemeansstrongerbasicity[26].
increasing acidity
0
14
CF3COOH
0.23
salicylic diclofenac
acid
3.99
2.88
phenobarbital acetaminophen
7.49
9.63
increasing basicity
0
caffeine benzocaine aminophenazone
0.60
2.39
14
debrisoquine
13.01
5.06
papaverine
6.39
amlodipine 9.26
propranolol 9.54
ephedrine 9.60
atropine
9.84
Figure1.5.ThepKascaleinaqueousmedium
Some examples for the most frequently occurring acidic and basic functional
groupsindrugsarelistedinTables1.1and1.2.
Equations1.1‐1.4showtheionizationequilibriaandtherelevantthermodynamic
ionization constants using general symbols: HA for acid, B for base, XH for
diproticampholytemolecule.
‐
+
HA  A +H K a 
pK a  pH  log

+
[HA]
[A  ]
[A  ][H  ]
HA 
BH  B+H K a 
(1.1a,1.1b)
(1.1c)
[B][H  ]
[BH ]
(1.2a,1.2b)
9
Chapter1
Table1.1.Someimportantacidicfunctionalgroupsindrugs
group
name
pKa
example(pKa)
CH3
sulphonicacid
0‐1
O
C
OH carboxyl
2‐7
O
enol
C C
OH N
N
CH3
OH O
C
C
C
N
H
N
S
CH3
O O
2‐6
N
N piroxicam
(2.33)
CH3
CH3
H3C
tetrazole
metamizole
benzoicacid
(3.98)
N
HOOC
O
N
N
CH3
OH
C
O H
N
N
HO3SCH2
O
S OH
O
valsartan
(4.8)
4‐5
N N
N N
H
O
O
S N C
O H
O
NH S
O
sulphonimide
5‐6
O
O
S N C N C4
H
O H
H3C
N‐aryl‐
‐sulphonamide
6‐8
O
N
S N
O H N
H2N
O
lactam
CH3
sulfadimidine
(7.49)
phenol
C
HN
7‐8
9‐11
CH3
O
OH
CH3
CH3
NH
tolbutamide
(5.3)
O
C
O
barbital(7.9)
NH
OH
HN
O C
CH3
acetaminophen
(9.63)
CH3
SH 10
thiol
8‐11
HS
O
N
COOH
captopril(9.8)
Physicochemicalprofilingindrugresearchanddevelopment
Table1.2.Someimportantbasicfunctionalgroupsindrugs
group
name
pKa
example(pKa)
NH
HN C
HN guanidine
N
13‐14
NH
NH2
debrisoquine(13.0)
OH
HO
aliphatic
primary
amine
NH 2 NH2
HO
noradrenaline(8.5)
OH
aliphatic
secondary
amine
NH
NH
8‐11
CH3
CH3
ephedrine(9.6)
CH3 O
C
N
H
CH3
aliphatic
tertiary
amine
N
CH3
N
CH3
lidocaine(7.9)
aromatic
primary
amine
NH
R aromatic
secondary
amine
NH2
O
C
O
H2N
CH3
benzocaine(2.4)
CH3
HN
2‐5
O
C
O
tetracaine(2.4)
CH3
N
CH3 Cl
N R
R' aromatic
tertiary
amine
N
N
CH3
N
CH3
chloropyramine(2.0)
11
Chapter1
pK a  pH  log
[BH  ]
[B]
(1.2c)
XH2  XH+H+ K a1 
XH  X ‐ +H + K a2 
pK a1  pH  log
[XH][H ]
[XH2 ]
(1.3a,1.3b)
[X‐ ][H ]
 XH (1.4a,1.4b)
[XH2 ]
[XH]
pK a2  pH  log  [XH]
[X ]
(1.3c,1.4c)
Incertainresearcharticles,preferenceisgiventotheuseoftheionizationrather
than the proton association process and the term protonation constant, Kp,
particularly in coordination chemistry [27]. The relationship between them is
reciprocal where Ka= 1/Kp, or pKa = log Kp. For a monoprotic compound this
relationshipisevident,butmaynotbeclearregardingmoleculeswithmorethan
one ionizable group. Below, we describe the ionization processes of a triprotic
compound (like amoxicillin) from both points of view: dissociation (molecule
releasestheproton)andassociation(moleculegainstheproton).
O
HOOC
N
H3C
H3C
N
H
S
OH
O
NH2
Dissociation
K p1
‐


X 2‐ +H+ 
 XH K a1
K p3
K a2 
(1.6a,1.6b)
K p3
(1.7a,1.7b)
+


XH2 +H+ 
 XH3 K a3
2‐
+


XH ‐ 
 X +H 1
Kp2
K a3 
1
K p1
pKa1=logKp3pKa2=logKp2pKa3=logKp1
12
(1.5a,1.5b)
K p2


XH‐ +H+ 
 XH 2 K a2
+



XH2 
 XH +H 1
Protonation
+


XH3+ 
 XH2 +H K a1 
log Kp1= 9.6
log Kp2= 7.4
log Kp3= 2.4
(1.8a,1.8b,1.8c)
(1.9a,1.9b,1.9c)
Physicochemicalprofilingindrugresearchanddevelopment
Ionizationmicroconstants
Theequilibriaabovecharacterizethedissociation/protonation ofthemoleculeat
the molecular level, so called macroscopic level, using ionization macroconstants.
Ionization macroconstants quantitate the overall acidity/basicity of the molecule,
butcannotbeassignedtoindividualprotonbindingsitesofmultiproticmolecules.
Ionizationmicroconstantsarethetermswhichdescribetheprotonbindingability
oftheindividualfunctionalgroupsandareusefulincalculatingthepH‐dependent
concentrationsofmicrospecies(namedmicrospeciation)[28].Inthepastdecade,
themicrospeciationofseveraldrugmoleculeswaspublished[e.g.29‐31].
The macroscopic and microscopic protonation scheme of a diprotic molecule
usingnorfloxacinasamodelisshowninFigure1.6.Forsimplicity,Kdenotesthe
protonation macroconstants and k is used for microconstants. The superscript
denotes the functional group is protonating in a given process, the subscript (if
any) shows the already protonated group and N and C refer to the piperazine
nitrogen and the carboxylate group, respectively. There are two possible alter‐
nativeroutesofprotonation.Fromthemostbasicanionicform(X‐)thecarboxy‐
late group first accepts a proton resulting in the chargeless (XHo) form, then a
secondary amine group protonates producing the cation (XH2+) (lower route).
The other pathway of protonation is conducted through the formation of a
zwitterion(XH±)duetotheprotonationof an aminogroup first.Thechargeless
and zwitterionic forms are chemically different microspecies (they bear the
proton on different binding sites) having the same stochiometric composition
(oneprotonisaccepted),sotheyareprotonationisomers.
O
-
F
k
H
H
N
N
k
C
N
O
CH3
H
-
F
N
N
+
N
O
COO
COO
N
F
N
CH3
k
C
-
K1
N
N
N
+
COOH
N
H
X
N
O
F
COOH
H
k
N
CH
3 H
N
C
CH3
XH
K2
XH2
+
ß2
Figure1.6.Theprotonationmacro‐andmicro‐equilibriaofnorfloxacin
13
Chapter1
Therelationshipsbetweenthemacro‐andmicroconstantsarethefollowing:
K1=kC+kNK1K2= k C k CN = k N kNC 1
K2

1
N
C
k

1
kNC
(1.10a,1.10b)
(1.10c)
Once macro‐ and microconstants are known, the mole fraction of each species
can readily be calculated and the pH‐dependent distribution of macro‐ and
microspecies can be constructed. Figure 1.7 shows the distribution of different
protonationformsofnorfloxacinagainstthepHandindicatesthepredominance
of the zwitterionic form over the chargeless microspecies. However, it is also
visiblethattheirconcentrationattheiso‐electricpointpHiscommensurableand
bothformsarepresentinasignificantamount.
100
80
%
60
40
20
0
2
4
6
8
10
12
pH
Figure1.7.Distributioncurveofthe4microspeciesofnorfloxacinasafunctionofpH
The microspeciation of a triprotic molecule [32,33] is more complicated, conta‐
ining8microspecies.Thetotalprotonationprocesscanbedepictedby12micro‐
constantsasdemonstratedincaseofamoxicillininFigure1.8.TheO,N,Csub‐or
superscripts of the k microconstant refer to the three proton binding sites,
namelyphenolate,amino,andcarboxylategroups,respectively.Therelationships
betweenthemacro‐andmicroconstantsarethefollowing:
14
K 1  k O  k N  k C (1.11)
K1K2  kOkON  kOkOC  kNkNC  kNkNO  kCkCO  kCkCN (1.12)
C
C
K 1 K 2 K 3  k O kON kO,N
 k N kNO kO,N
 ........ (1.13)
Physicochemicalprofilingindrugresearchanddevelopment
The theory and practice of proton microspeciation based on NMR‐pH titration
anddataintheliteratureoncompletemicrospeciationofsmallligandsincluding
drugshaverecentlybeensurveyed[34].
HO
HO
NH2
O
N
O
O
-
-
OOC
S
CH3
H3C
O
C
NH2
C
k N ,O
HO
HO
N
H3 C
NH2
NH3
O
N
k
NH
O
OOC
O
kN
kO
O
-
N
CH3
H3 C
NH
O
S
OOC
k
-
NH3
O
NH
O
-
N
kO
O
NH
O
N
S
-
CH3
OOC
-
N
S
CH3
H3 C
NH
O
-
O
O
NH
N
S
H3 C
CH3
O
k
C
kN
NH3
NH2
O
N
HOOC
S
H3C
XH
-
O
N
NH
O
2-
NH3
k N,C
C
X
HOOC
CH3
H3 C
O
O
S
HOOC
O
kC
N
k O ,C
kC
NH
O
N
CH3
HOOC
H3 C
XH2
S
CH3
XH3
+
Figure1.8.Protonationmacro‐andmicro‐equilibriaoftriproticamoxicillin
Temperatureandionicstrength
The ionization constant as a thermodynamic parameter is temperature‐
dependent.FortheprecisedeterminationofpKa,experimentsmustbeconducted
undercontrolledconstanttemperature.Inpractice,thecommonreferencevalue
is 25 °C and only few data are available measured at 37 °C. The change in pKa
uponanincreaseoftemperaturefrom25°Cuptothephysiologicaltemperature
of37°Cisdependentonthegiven molecule.Generallythechangeinthe pKaof
acids is less, while bases are more sensitive to temperature change [26]. The
approximate average value of temperature dependence is known as δpKa/δT:
0.02‐0.03,whichmeans0.24‐0.36ΔpKavaluesbetween25and37°C.IfthepKaof
acompoundfallsintothepHrange1.5‐8(thepHgradientpresentinthehuman
gastrointestinaltract),thenevenarelativelysmalldifferencemayleadtopoorin
15
Chapter1
vitro‐in vivo correlations. For a better interpretation of the cellular transport
mechanismofsuchmolecules,thebiorelevantpKavalueisparticularlyuseful.A
prediction method for this value based on a 2D structure and pKa at 25 °C was
proposedveryrecently[35].
The ionic strength of the medium also affects the pKa value. It is common to
measure at constant ionic medium, generally at I = 0.15 M adjusted by KCl or
NaCl corresponding to the physiological level. Frequently, a different ionic
mediumisusedordatacalculatedtozeroionicstrengthusingtheDebye‐Hückel
theoryarealsopublished,thusitisalwaysnecessarytoreporttheionicstrength
andtemperatureofapKameasurement.
ImportanceofpKainmedicinalchemistry
ThedegreeofionizationatagivenpHcanbecalculatedoncethepKaisknown.As
aruleofthumb,atpH=pKa50%ofthecompoundisionizedand50%isinthe
unionizedform,whileatpH=pKa±2predominanceofonespeciesbecomes99
%.Forexample,anacidispresentin99%atpH=pKa‐2asunionized(HA)and
atpH=pKa+2asionized(A−)(theoppositecaseappliestoabase).
The ionization state determines the transport properties, thus its precise
calculation allows the estimation of ADME features. With the knowledge of the
pKa value, the proportion of the transport form can be calculated at any
physiologically important pH values. Regarding ampholyte compounds, the pKa
valuesareusefultocalculatetheiso‐electricpointorthepHatwhichamolecule
has the lowest solubility and highest lipophilicity. Since solubility, lipophilicity,
andpermeabilityarepH‐dependentproperties,thepKavalueofanewmolecule
mustbedeterminedinadvancetothelogS,logPandpermeabilitymeasurement.
Ionic interactions play a fundamental role in the receptor binding of ionizable
molecules. An ionic bond is the strongest non‐covalent binding type. The
electrostatic attraction of opposite charges directs the molecule to the receptor
surfaceandelectrostaticcomplementaritywiththereceptor isaprerequisiteof
anydrugaction.
Antiarrhythmicdrugs(classI:Na+‐channelantagonists)serveasagoodexample
ofhowpKaaffectsdrugaction.ThesedrugsareweakbaseswithmosthavingpKa
values ranging from 7.5 ‐ 9.5. At the physiological pH of 7.4 they exist in an
equilibrium mixture consisting of both the free base (B) and protonated (BH+)
cationicform.IncompoundswiththepKa>9(likeprocainamide,mexiletine,pro‐
pafenone),thepresenceofthereceptorformexceeds90%whichisfavorablefor
thebindingtothesodium‐channel.However,forcompoundsinwhichthepKa<8
(likequinidine,lidocaine)thisratioismuchlessfavorable(Table1.3).Lidocaine
(pKa = 7.96) has a stronger electrophysiologic effect in ischemic than normal
myocardialtissue.Thispotentiationhas,inpart,beenattributedtotheincrease
in H+ concentration (lower pH) within the ischemic areas of the heart. Acidosis
increasestheportionofreceptorformofthedrug(Table1.3)andconsequently
theproportionofNa+‐channelsoccupiedbytheBH+oflidocaine[36].
16
Physicochemicalprofilingindrugresearchanddevelopment
Table1.3.IonizationstateofantiarrhythmicclassIdrugsatnormalandischemictissue
compound
procainamide
pH=7.4(normaltissue)
pH=6.4(ischemictissue)
BH+,%
B,%
BH+,%
B,%
98.4
1.6
99.9
0.1
mexiletin
98.2
1.8
99.8
0.2
quinidine
76.0
24.0
99.0
1.0
lidocaine
78.4
21.6
97.3
2.7
1.2.2.2. Solubility(logS)
Solubility is a molecular property which determines the maximal concentration
ofasoluteinagivensolvent.Theaqueoussolubilitydependsonthepolarityofa
molecule and varies with the pH for ionizable compounds. Solubility can be
described by different parameters and a vast variety of terms and symbols are
usedtoexpressthesolubilitydataofcompounds,hencebelowwesummarizethe
mostcommonbasicdefinitions.
Definitions,terms
Equilibrium(orthermodynamic)solubility(S)istheconcentrationofacompound
inasaturatedsolutionwhenasolidispresentandthesolutionandsolidareat
equilibrium. This value is constant at a given pressure and temperature and
characteristicforagivencompound.
Forionizablemolecules,furthertermsaredistinguished.Intrinsicsolubility(S0)is
theequilibriumsolubilityofafreeacid(HA)orfreebase(B)formofanionizable
compound at a pH where it is fully unionized. With respect to ampholytes, this
referstotheneutral(chargeless)form(XH)whichexistsattheiso‐electric(i.e.)
point pH. Effective solubility (SpH) is the equilibrium solubility of an ionizable
compoundatapHwherebothunionizedandionizedformsarepresent.Itisalso
denotedasapparent(ortotal)solubilityanddefinedataparticularpHasthesum
oftheconcentrationsofallcompoundspeciesdissolvedinanaqueoussolution.
Thesolubilityofsaltformofanionizablecompound(Ssalt)canbederivedfromthe
solubilityproduct(Ksp).Foramonoproticacidorbase:
S salt  K sp
where, Ksp = [A−] [Y+] for an acid and Ksp = [BH+] [X−] for a base, Y+ and X−
representthecounter‐ioninthesalt.
Recently,anewtermcalledkineticsolubility(SAPP)wasintroducedinearlydrug
discovery.Itistheconcentration of asolutionof anexaminedcompound atthe
moment when the first precipitation of the solid is observed in an experiment
whereasmallvolumeof10‐20μg/mldimethylsulfoxide(DMSO)stocksolution
is added to aqueous buffer. This parameter is not a thermodynamic physico‐
chemical constant because the system does not reach an equilibrium state.
Generally,SAPPishigherthantheequilibriumsolubilitysincethereisnoneedto
overcome the crystal lattice forces by aqueous solvent once the compound has
17
Chapter1
been dissolved in DMSO. Kinetic solubility data are mainly used for ranking the
moleculesintheearlystagesofdiscoveryandcannotreplacethedetermination
ofthetrueequilibriumconstantlaterinthedevelopmentphase.
Theabovesolubilityparameterscanbeexpressedinvariousconcentrationterms
like:g/100ml;g/ml;mg/ml;μg/mlormol/L(M);mmol/L(mM); μmol/L (μM),
etc.Forbettercomparability,thelogarithmofsolubilityterm(logS)isfrequently
usedandcanbeobtainedfromMorμMconcentration.Preferenceforthe–logS
term is found in the literature in order to avoid negative numbers for low
solubility compounds. However, it may be somewhat confusing because the
aforementionedtermyieldshighervaluesmeaninglowersolubility.
Solubility is affected by many factors, such as temperature, pressure, pH, ionic
strength of aqueous media, purity of a sample, crystal form, particle size, poly‐
morphism,etc.Theeffectofthesefactorshavebeencomprehensivelydiscussed
inclassic[37,38]andnewbooks[11,18].Here,wefocusonly onthepHdepen‐
dencyofsolubility.
Solubility‐pHprofile
ThesolubilityofionizablecompoundsvarieswiththepH.Theyaremoresoluble
in the charged than in the unionized form. When a molecule exists only in the
monomer state, its pH‐dependent equilibrium solubility is derived from the
Henderson‐Hasselbalch (HH) equations (Equations 1.1c‐1.4c). The HH relation‐
shipforamonovalentacid,base,and(diprotic)ampholytemoleculecanbederi‐
ved from solubility and ionization equilibria as follows where, by convention
[HA(s)]=[B(s)]=[XH(s)]=1, and [A−], [BH+], [X−], [XH2+] are expressed using
Equations1.1b,1.2b,1.3b.
acid:
HA(s)⇌HA
S0 
(1.14)
[HA]
 HA [HA(s)]
(1.15)
S pH =  A   + HA  S pH 
(1.16)
K a [HA]
 HA  [H ]
(1.17)
 K

S pH  HA   a  1   [H ] 

S pH  S0 10(‐pK a  pH)  1

(1.18)
log S pH  log S0  log(1  10(pH‐pK a ) ) 18
(1.19)
(1.20)
Physicochemicalprofilingindrugresearchanddevelopment
base:
B(S)  B S0 
[B]
 B [B(s)]
(1.21)
(1.22)
...
log S pH  log S0  log(1  10(pK a ‐pH) ) (1.23)
diproticampholyte:
HX(s)  HX S0 
[XH]
  XH [XH(s)]
(1.24)
(1.25)
S pH =  X   +  XH +  XH2+  ...
log S pH  log S0  log(1  10(pK a1 ‐pH)  10(pH‐pK a2 ) ) (1.26)
Figure1.9showsthecharacteristicsolubility‐pHprofile(aplotoflogSpHvs.pH)
foranacid(a),base(b),anddiproticampholyte(c).
Figure1.9.Solubility‐pHprofileof(a)anacid,(b)abaseandc)adiproticampholyte
19
Chapter1
The HH relationship can be used to predict the pH‐dependent equilibrium
solubilityofdrugswhenthepKaandlogS0valuesofacompoundareknown.Itis
a frequent practice in drug research to convert the experimentally measured
intrinsicsolubilityvaluetoequilibriumsolubilityataphysiologicalrelevantpHin
ordertoestimateitsexpectedbehavior.
The validity of the HH relationship has been widely investigated and certain
deviations were found [39,40]. They were interpreted with the influence of
different molecular interactions such as aggregation and micelle formation
[39,41]. Recently, a revisit of the HH relationship concerning organic bases
confirmedthevalidityprovidedifhighlyprecisepKaandlogS0valueswereused
foritsgeneration[42](seealsoSection1.4.2).
ImportanceoflogSinmedicinalchemistry
The aqueous solubility of compounds receives considerable attention in drug
development, because this is a key molecular property for the gastrointestinal
absorption of orally administered drugs. Further on, in biological activity tests
compoundsmustbeinsolutionotherwisefalse,erroneousdatacanbeobtained.
Lowsolubilityisdetrimentalfrombothpharmacokineticandpharmacodynamic
points of view. Determination of aqueous solubility is an inevitable part of
physicochemical profiling in drug research. Its importance has grown since the
BiopharmaceuticalClassificationSystem(BCS)wasfirstproposedbyG.Amidon
in 1995 [43]. This classification uses four classes to categorize drugs based on
their solubility and intestinal permeability (class 1: high solubility + high per‐
meability;class2:lowsolubility+highpermeability;class3:highsolubility+low
permeability; class 4: low solubility + low permeability). For class 1 molecules,
the rate‐limiting factor of intestinal absorption is the rate of dissolution, low
solubilityinclass2molecules,whilelowpermeabilityinclass3israte‐limiting.
In class 4, both properties are unfavorable for oral administration, and no in
vitro‐invivocorrelationcanbeexpected.
To improve the in vitro‐in vivo correlation, the measurement of solubility is
recommended for biomimetic media as well. There is growing evidence that in
theintestine,thepresenceofbileacidsandothercomponentssuchaslipidscan
alter (usually increase) the intrinsic solubility of (lipophilic) compounds. Two
physiologically relevant media developed by Dressmann et al. [44] are used.
These are the fasted‐state simulated intestinal fluid (FaSSIF) and the fed‐state
simulated intestinal fluid (FeSSIF) having pH 6.5 and 5.0, respectively, and
containdifferentamountsofsodiumtaurocholate,lecithineandsalts[44].
1.2.2.3. Lipophilicity(logP)
Themorefundamentalpropertygoverningthefateofadruginthebodyisundo‐
ubtedlythelipophilicity.Thismolecularpropertyrepresentstheaffinityofamo‐
leculeforalipophilicenvironment.Itismostcommonlydescribedbythelogari‐
thmofpartitioncoefficient(logP)betweentwoimmisciblesolvents,oneisanor‐
ganicapolar(e.g.octanol)andtheotheranaqueouspolar(buffersolution)[45].
20
Physicochemicalprofilingindrugresearchanddevelopment
BesidesP,othersymbolshavebeenusedintheliteraturesuchasPow,Kow,Kp,PC,
etc.,however,weusetheterminologywidelyacceptedinmedicinalchemistry.
Two types of partition parameters are distinguished: the true partition coef‐
ficient (P) and the distribution coefficient (D or in older literature Papp). Their
definitionandrelationshiparebrieflysummarizedbelow.
Definitions,terms
Thetruepartitioncoefficient(accordingtotheNernstlaw)referstothepartition
of a single electrical species, and is expressed as an equilibrium concentration
ratioofthesamemolecularforminbothphasesofthesolventsystem.Thisvalue
is constant at a given temperature and pressure, independent of the pH and
characteristicforthemolecule.Itcanbederivedfortheneutral,monomericform
ofacompound(logPN)(Equation1.27)andtheoreticallycanbealsodefinedfor
the partition of an ionic form (log PI) (Equation 1.28), but later the value has
ordersofmagnitudelowerandinmostofcasescanbepracticallyneglected.
PN 
[unionizedform]octanol
[unionizedform]water
PN 
[HA]o
[HA]w
PI 
PN 
(1.27)
[B]o
[XH]o
N
P 
[B]w
[XH]w
(1.27a-c)
[chargedspecies]octanol
[chargedspecies]water
(1.28)
[A‐ ]o
[BH ]o
[X‐ ]o
[XH2 ]o
I
I
I
P  ‐ P 
P  ‐ P 
[A ]w
[BH ]w
[X ]w
[XH2 ]w
I
(1.28a-d)
The distribution coefficient of an ionizable compound refers to all species that
arepresentinthesolution(Equation1.29).SinceitisapH‐dependentterm,the
pHmustbespecifiedasDpH.
DpH 
[unionized  ionizedspecies]octanol
[unionized  ionizedspecies]water
(1.29)
For monoprotic acid and base:
DpH 
[HA]o  [A ]o
[HA]w  [A ]w
pH
D 
[B]o [BH ]o
[B]w [BH ]w
(1.29a‐b)
Fordiproticampholyte:
DpH 
[X ]o [XH]o [XH2 ]o
[X ]w [XH]w [XH2 ]w
(1.29c)
21
Chapter1
with the assumption that the concentration of the ionic forms in the organic
phase is much less than that of the neutral forms (e.g., [A−]o << [HA]o and
[BH+]o<< [B]o, etc.) and upon substituting the aqueous equilibrium concentra‐
tionsfromEquations1.1b,1.2b,and1.3b,therelationshipsbetweenPandDcan
be obtained. For simple molecules, these relationships are given below
(Equations 1.30a‐c) while interactions between more complicated multiprotic
compoundscanbefoundintheliterature[46].
Formonoproticacid:
log P N  log DpH  log(1  10(pH pK a ) ) (1.30a)
Formonoproticbase:
log P N  log DpH  log(1  10(pK a pH) ) (1.30b)
Fordiproticampholyte:
log P N  log DpH  log(1  10(pK a1 pH)  10(pH‐pK a2 ) ) (1.30c)
Partitionmicroconstants
Similarlytoionizationmicroconstants,micro‐logP(denotedaslogp)ofagiven
microspecies of multiprotic compounds can also be defined [46]. This has
particular significance in the case of ampholyte compounds where the most
lipophilicspecies,theneutral(XH)form,isacompositefromzwitterionic(XH±)
andchargeless(XH0)microspecies.Iftheyarepresentinsolutioninacommen‐
surableamount(e.g.,norfloxacininFigure1.7)thentheexclusivepartitioningof
thechargelessformcanbeexpectedintothelipophilicphaseandmicro‐logPof
XH0 microspecies may be the relevant lipophilicity parameter. Its calculation
requires knowledge of the log D at iso‐electric pH value, log Di.e.pH, and the
protonationmicroconstants(kC, kCN , k NC )aspreviouslypublished[47].


kN
1
log p0  log D i.e.pH  log  1  C +  CC  kCN H+   

k H  kN


(1.31)
Lipophilicity‐pHprofile
The plot of log DpH against the pH (lipophilicity‐pH profile) of a compound
(Equations1.30a‐c) can be derived from the HH relationships (Equations 1.1c‐
1.4c),providedthatthereisnoion‐pairpartitioninvolvedintheprocess.Ifsuch
ion‐pairpartitionexists,theprofilesshowaplateauatlogDofvalues3‐4orders
lower(foracidsathighpH,forbasesatlowpH)thanasindicatedinFigure1.10.
Thelipophilicity‐pHprofilesareusefultoestimatetheeffectivelipophilicityofa
compound at physiologically relevant pH values and widely used in medicinal
chemistry.
22
Physicochemicalprofilingindrugresearchanddevelopment
Figure1.10.Lipophilicity‐pHprofileof(a)anacid,(b)abase,
and(c)adiproticampholyte
SolventsystemsforlogP
The widely accepted reference solvent system for log P measurement is
octanol/waterproposedfirstbyHansch[45].Inthissystem,thetwophasesare
isotropic.Abuffersolutionservesastheaqueousphaseandn‐octanolisusedasa
typical H‐bond donor and acceptor organic solvent. This system is thought to
modeltheessentialpropertiesofgeneralbiologicalmembranes.Thestructureof
water‐saturated octanol became better understood in the ‘90s [48]. Inverted
micellaraggregatesareformedwherewaterclustersaresurroundedbyabout16
molecules of octanol, with the polar OH groups pointing to the clusters and
intertwined in a hydrogen‐bonding network. The aliphatic tails form a hydro‐
carbon region with properties not too different from the hydrocarbon core of
bilayers.Obviously,theoctanol/watersystemcannotbeauniversalmodelforall
types of membranes. In the past two decades, partition solvents other than
octanol have been explored. Leahy et al. [49] proposed the “critical quartet”
system consisting of octanol/water, chloroform/water, alkane/water and pro‐
pyleneglycoldipelargonate(PGDP)/waterforthegeneralmodelingofmembra‐
nes.Later,1,2‐dichloroethane(DCE)andcyclohexanewerefoundusefulorganic
solvents.OncelogPhasbeenmeasuredbothinalkane/waterandoctanol/water
systems,theΔlogP(logPoctanol–logPalkane)canbecalculated,andusedasasimple
parameterfortheH‐bondformationabilityofacompound.
Recently, anisotropic systems such as liposomes (vesicles formed from
phospholipidbilayers)wereincreasinglyusedtomodelmembranepartitioning.
23
Chapter1
Liposome/water log P values are considered as log Pmem (membrane partition).
AnaccumulationoflogPmemdatashowasignificantlyhigherpartitioningofionic
forms. Generally, charged species partition into membranes about 100 times
more strongly than into octanol. The theory and practice of liposome/water
lipophilicitywerereviewed[5,50,51].
ImportanceoflogPinmedicinalchemistry
The log P is the oldest and most traditional physicochemical parameter used in
medicinal chemistry. Lipophilicity is implicated in numerous biological events
(such as transport, receptor binding via hydrophobic interactions, metabolic
processes, storage in fat tissues, etc.). The log P value – concerning its infor‐
mationcontent‐ismuchmorethanasimplenumber,becausethesamemolecul‐
ar interactions which exist between the compound and the biological environ‐
ment results in this value. At the same time, log P is very easy to handle by
chemistsforcomparisonofmoleculeswithdifferentlipophilicityandestimating
theexpectedtransportbehaviorinthebody.
Among the properties suggested by Lipinski, (known as “rule of 5”) one of the
criteria for drug‐likeness is that log P should be below 5 [52]. It seems to be a
reasonableconceptsince90%ofmarketeddrugshavealogPvalueintherange
of 0 – 5 (see Figure 1.11). From hydrophilic compounds (log P < 0) good
solubility,butpoorabsorptionfromtheGItractcanbeexpectedexceptforthose
whichhaveactivetransport(suchasforexampleascorbicacid).Compoundswith
moderate lipophilicity (log P between 0 and 3) are optimal for oral
administration due to a good balance of solubility and permeability. For good
blood‐brainbarrier(BBB)penetration,theoptimallogPvalueisabout2.
log P scale
…
…
-2
0
2
4
8
6
drugs 90%
HO
OH
O
HO
O
OH
H 3C
+ CH 3
N
Br-
I
S
O
O
OH
O
O
N
Cl
CH 3
N
CH 3
N
CH
3
CH 3
I
O
CH 3
ascorbic acid
methylhomatropine -bromide
chlorpromazine
amiodarone
log P:
P: -1.85
absorbs by active
transport
log P: -1.68
no absorption
no BBB penetration
log P: 5.34
good oral absorption
good BBB penetration
log P: 7.57
storage
t1/2: 25-30 days
Figure1.11.ThelogPscaleofdrugs
Highly lipophilic compounds (logP>5) are sparingly soluble in aqueous
compartments,tendtoaccumulateinlipoidalpartsandarealsomoresensitiveto
24
Physicochemicalprofilingindrugresearchanddevelopment
metabolism. Extremely high lipophilicity may lead to strange pharmacokinetics,
for example, amiodarone has log P=7.37 and half‐life t1/2: 25‐30 days(!)
(Figure1.11).First,in1987Hanschcalledattentiontothedangerofexceedingly
high lipophilic drug candidates and proposed the “minimal hydrophobicity”
concept for the design of new compounds [53]. Since then, the unfavorable
tendencyofhighlylipophilicdrugproductionhasnotstopped,asnewmolecules
indrugresearcharegettingmorelipophilicandlesswater‐soluble[54].
1.3.
METHODSFORPHYSICOCHEMICALPROFILING
Demands set up to the methods for physicochemical profiling are different in
variousphasesofdrugresearch.Inthediscoveryphase,thedrasticallyincreased
numberofNCEsproducedbycombinatorialchemistryrequireshighthroughput
(HT), material saving, automated approaches, while less emphasis is placed on
precision. A method for physicochemical profiling is considered HT when its
capacity exceeds the measurement of 50 compounds/day [7]. Later, in the
development phase reliable, precise data are needed which is why accuracy is
moreimportantandnotthespeedofthemethod.
Thissubchapterisdedicatedtoexperimentalmethodsusedforthemeasurement
of pKa, log P, and log S values and comparison of their capacity, accuracy, time,
andmaterialdemand(seeTables1.4‐1.6).Wefocusheremainlyonthepractical
aspects of their application, while the detailed theoretical background of the
methodsisoutofthescopeofthisreview.Forthispurpose,excellentbasicbooks
arerecommendedtoreaders[11,18,26,37].
1.3.1. pKadetermination
PotentiometryandUVspectroscopyarethecommonlyusedstandardmethodsof
pKa determination. Due to its simplicity and precision, potentiometry is the
methodofchoiceoncetheaqueoussolubilityofacompoundreachesaminimum
of0.5mMconcentrationintheentirepHrange ofthetitration.Forlesssoluble
compounds, a good alternative tool is the UV/pH titration provided that the
moleculehasapH‐dependentUVspectrum.Inthismethod,itisgenerallyenough
if the compound dissolves in a concentration of 10‐500 μM depending on its
molar absorptivity, ε. Both potentiometric and UV/pH titration methods are
stronglysupportedcommercially,andtheavailableautomatedinstrumentssuch
astheGLpKaanditsfollowuptheSiriusT3automatedanalyzers(SiriusUK)are
widelyused.Intherecentyears,capillaryelectrophoresis(CE)hasprovedtobea
verypowerfulpKadetermination method,being moresensitiveandless sample
consuming[11,16].SomeothermethodssuchasNMR/pHtitration[55],CD/pH
titration [56], and chromatographic technique [57] have also been applied for
specialcases,butsofarhavenotbecomeroutinetechniques.
25
Chapter1
1.3.1.1. Potentiometricmethod
Procedure.Inpotentiometrictitration,thepHofa1‐5mMsolutionofasampleis
preciously measured with a carefully standardized combined glass electrode
upon addition of small volumes of a strong acid (e.g. HCl) or base (e.g. KOH)
volumetricsolution.Themeasurementisperformedinastirringsolution,under
an inert gas atmosphere (argon or nitrogen) while the ionic strength of the
solutioniskeptconstantusinganinorganicsalt(e.g.0.15MKCl),andthetitration
cellisthermostatedusuallyat25.0±0.1°C.Typicalsamplevolumefortitrationis
5‐15ml,butmeasurementinaslessas1mlsolutionhasbeenreported[58].The
concentration of the titrant is generally 0.5 M in order to avoid considerable
dilution upon titration. The potentiometric titration can be used as a direct
approach for pKa measurement, when the tested compound is a (relatively)
strongacid/basetoproduceenoughpotentialchange(bigjump)inthetitration
curve. Otherwise, the “Calvin‐Δml” difference‐titration is a useful and widely
appliedmethod.Here,thepKavalueisobtainedfromthedifferencebetweenthe
titration curve of a tested compound and a “blank” titration (see below). This
approachisabuiltinfunctioninpKaanalyzers.
Calculation. The pKa value can be calculated according to the HH equations
(Equations1.1c‐1.4c).ThepHismeasuredandthetermlog([protonated]/[non‐
protonated]) is obtained from the mass balance of the titration data. In
automatedanalyzersbuiltinprograms(e.g.,Refinement‐ProTM)calculatethepKa.
First,thetitrationcurveisconvertedtotheBjerrumplot(theaveragenumberof
boundprotons/molecule, n vs.pH),wherethepKavalueisequaltothepHat n =
0.5(foramultiproticcompound:secondpKaat n =1.5,thirdat n =2.5,etc.).The
obtained raw values are then further refined by a nonlinear least squares
method. The adjustable parameters are the concentration of the material,
acid/baseerrorofpHmeasurement,carbondioxidecontent,etc.
Accuracy. This method with the above experimental parameters allows the
measurement of precise pKa values in a range from 2 to 12 with a standard
deviation SD = ± 0.01‐0.03. By using a glass electrode of excellent quality,
performing proper electrode calibration, excluding the presence of ambient
carbondioxideasmuchaspossible,andaccuratelydispensingverysmalltitrant
volumes (0.01 ml or even smaller) potentiometry in aqueous solution can be
appliedtoaconcentrationaslowas0.1mM(accordingtosomeauthorsaslowas
0.01mM).Ofcourse,theaccuracyandreproducibilityoftitrationsinsuchdiluted
solutionsismuchless(SD=±0.10‐0.15).Similarly,theprecisionofthemeasure‐
mentdecreasesoutofthepHrangeof2‐12.
Calibration. Electrode calibration is a fundamental step in pH‐metric pKa
determination.Astandardized“Four‐parameterprocedure”developedbyAvdeef
et al. [59] is widely used. A known concentration of HCl is titrated with KOH
(frompH1.8to12.2)understandardexperimentalconditions(seeabove).Data
from this “blank” titration are used to convert the operational pH scale to the
concentrationscale(pcH=−log[H+])byamulti‐parametricequation.
26
Physicochemicalprofilingindrugresearchanddevelopment
pH=α+SpcH+jH[H+]+jOHKw/[H+]
(1.32)
The parameters are determined by a weighted nonlinear least‐squares
procedure.Theinterceptparameterαinaqueoussolutionmainlycorrespondsto
thenegativelogarithmoftheactivitycoefficientofH+attheworkingtemperature
and ionic strength. The jH term corrects pH readings for the nonlinear pH
response due to the liquid junction and asymmetric potentials in moderately
acidicsolutions,whilethejOHtermcorrectsthehigh‐pHnonlineareffect.FactorS
accounts for the fact that a particular electrode may not have 100% Nernstian‐
slope and Kw is the ionization constant of water. Typical aqueous values of the
adjustableparametersat25°Cand0.15Mionicstrengthare:α=0.08±0.01,S=
1.001±0.001,jH=1.0±0.2,andjOH=−0.6±0.2.
Advantages/drawbacks. Potentiometry is a simple, fast, and precise method for
pKa determination. The smallest practical volume of sample solution is about
5ml.Thisrequires1.5mgofsampleforacompoundwithMw300toachievethe
1mM concentration which is ideal for titration. For reliable pKa, 2–3 parallel
measurements are necessary, so the sample consumption reaches 3–4.5 mg. A
titration between pH 2–12 typically takes 20‐40 min to perform. With an
automated instrument (e.g. GLpKa) 30‐40 titrations could be performed during
one 24‐h day [60]. So, the maximum capacity is about 10‐12 compounds/day.
Thisisarelativelylowthroughput.Themainlimitationoftheapplicationofthis
technique is the poor solubility of compounds. In such cases, the co‐solvent
method can be applied (see Section 1.3.1.4). Further on, it is difficult to handle
impureorunstablecompounds(e.g.,certainesters,diphenols,etc.).
1.3.1.2. UV/pHtitration
Procedure. In spectrophotometric pKa determination method the change in the
UVspectrumuponionizationisregistered.SuchapHdependentUV‐spectrumis
obtained if the ionizable group is near to the chromophore of the molecule. In
traditional UV/pH titration two aliquots of typically 10‐50 μM solutions of a
samplearepreparedineither0.01(or0.001)MHClor0.01(or0.001)MNaOH,
with the total ionic strength of 0.15 M. In one solution the molecule is fully
ionized while in the other fully unionized. By mixing the two stock solutions
underprecisepHcontrol,5‐6solutionsarepreparedinarathernarrowpHrange
(± 0.6 unit) around the expected pKa. Their absorbance is measured at a
wavelength where the difference in the absorbance between the ionized and
unionized form is the largest. Recently, this time‐consuming process has been
automated (GLpKa with a D‐PAS attachment). In a titration cell, the solution of
thesampleistitratedacrossapHrangethatincludesthepKavalue(s)andmulti‐
wavelength UV spectra registered at each pH with the help of a fiber optics dip
probeimmersedintothetitrationcell[60,61].
Calculation. In traditional UV/pH titration, the pKa value can be calculated from
thepHofthesolutionandtheabsorptiondatameasuredatasinglewavelength
using the HH Equations 1.1c‐1.4c. The pKa of a compound is obtained as an
27
Chapter1
average value calculated from the solution series. This method is applicable for
thedeterminationofasinglepKa,ormultiplepKavaluesiftheyarewellseparated
(>1.5 pH units). In the D‐PAS technique, target factor analysis (TFA) is used to
deduce the pKa value(s) of a sample from an absorbance matrix [60]. This
techniqueisabletohandlemultiproticmoleculeswithoverlappingprotonation.
Accuracy.TheprecisionofpKadeterminationbytraditionalUV/pHtitrationdoes
not reach that of pH‐metry, where the standard deviation can vary between
± 0.05‐0.10. However, according to a recent validation study, the D‐PAS tech‐
niquewithaSD=±0.02hassimilarprecisiontopotentiometry[62].
Advantages/drawbacks. The spectrophotometric method is usually more
sensitive than potentiometry. The measurements can be performed at lower
sampleconcentrationallowingthepKadeterminationoflesssolublecompounds
directly in aqueous medium, while for water‐insoluble materials the co‐solvent
methodcanbeeasilyapplied.TheD‐PASisafasttechnique,onetitrationtakesup
30 min and is sample conserving, usually 1‐2 mg of sample is enough for 3
parallel measurements. One limitation of spectrophotometry is that if the
distance between the ionization and the chromophore center is greater than
three sigma bonds then the pH‐dependent spectral shift will be too small for
measurement. Another limitation is if the absorption maxima of the compound
occurs at a low wavelength (< 230 nm) then background noise disruption
increases considerably. Traditional UV/pH titration is a very slow, time‐
consuming process, while the capacity of the D‐PAS technique is similar to
potentiometry(10‐12compounds/day).UV/pHtitrationwasusedfordetermina‐
tionofmicroconstantsinseveralcases(e.g.repaglinide[63],moxifloxacin[64])
when the shift in the UV spectrum is due to the ionization of a given functional
group.
1.3.1.3. Othermethods
NMR/pHtitration.NMR/pHtitrationcanalsobeusedforpKameasurementbased
onthefactthatthechemicalshiftofNMR‐activenucleiisgoverned(amongother
factors)bytheprotonationstateofionizablegroups.Sinceprotonationdecreases
the local electron density, a selected nucleus in the vicinity of the ionizable site
exhibitsadifferentshiftintheionizedandunionizedstates.Aplotδobsvs.pHhas
asigmoidalshapewithaninflectionpointatpH=pKa.
Generally, NMR/pH titrations have been carried out in aqueous solutions using
D2Oasasolvent.AlthoughglasselectrodesoperateproperlyinD2O,acorrection
factor of 0.40 has to be added to the measured pH to get the true pD value. To
avoidthiscorrection,NMR/pHtitrationmaybeconductedinasolventmixtureof
H2O/D2O(90/10v/v)andthewaterpeakhastobesuppressedbyanappropriate
method. Frequently, the whole titration is performed in a single NMR‐tube and
the pH is measured with a long, thin glass electrode. This method has been
extended for the measurement of low pKa values (between 0 and 2), where po‐
tentiometry is no longer applicable. Since at such low pH a glass electrode has
28
Physicochemicalprofilingindrugresearchanddevelopment
significantacidityerror,dichloroaceticacidwasproposedasanNMR“indicator
molecule”forinsitumonitoringofthepHinstrongacidicsolutions[55].ThepKa
valuesofindividualgroupsoflargebiopolymershavebeenreportedasmeasured
byNMRtechnique[65].
The main advantage of this technique compared to potentiometry is the capa‐
bility of selective monitoring of ionization of a given functional group in multi‐
protic molecules with overlapping protonation. Thus, this methodology has
become the chief approach of microspeciation as reviewed recently [34]. The
acid/base profiling of imatinib [66] and cetirizine [33], measured by NMR/pH
titrationwasreported.
Capillary Electrophoresis (CE). The application of CE for pKa determination has
been intensively growing in the past decade as reviewed [7,16,67,68]. The
method utilizes the change in electrophoretic mobility of a compound with
changeinpH.Theeffectivemobility(μeff)ismeasuredatvariouspHvaluesand
pKaisobtainedfromtheplotofμeffvs.pH.Theexperimentalconditionseffectthe
pKa determination such as buffer type and ionic strength, applied voltage,
detectionmethod,etc.arediscussedasdetailedbyHenchozetal.[16].
Inthistechnique,thesampleconsumptionissmall(ng),andimpuresamplescan
be handled due to the separation upon the analysis. It is rather universal, since
different detection methods can be coupled to CE [69]. The precision is good
enoughandagreementwithothermethodsisacceptable,about±0.2pKaunitsin
arangefrom2to10,butcanbemuchweaker(±0.5)outofthispKarange.The
method is sensitive for several factors, among them temperature which is
cardinal.
Today, CE is a good tool for high throughput pKa measurement. The instru‐
mentationisfullyautomatedusingamultiplex96‐channelCEwithUVdetection
(CombiSep,Ames,USA)andmorethan150samples/daycanbemeasured[70].
Spectral Gradient Analysis (SGA). To further increase the throughput of
physicochemicalprofiling,arapidpKadeterminationmethodwasdevelopedand
reported first as “pH‐gradient titration” [71]. Later, after the launch of a com‐
mercialinstrument(ProfilerSGA,Sirius)itisreferredtointheliteratureasthe
SGA method. In this technique, a pH gradient flow – very linear in time – is
created by mixing appropriate acidic and basic buffers. The sample is injected
into this pH gradient flow which passes through a diode array UV spectro‐
photometerandthespectraareregistered.ThepHisnotmeasuredbutestimated
fromthetimeelapsedsincethestartofthegradientgeneration.ThepKavalues
aredeterminedfromchangesinabsorptionasafunctionofpH.Thecalculationis
basedon eitherthefirst derivativeplotoftheabsorptionspectrumforsamples
withonlyasinglepKa(orwellseparatedpKavalues)orontheTFAapproachfor
compounds with weak spectral change or overlapping ionization [60]. The
precisionofthemethodisevidentlylowerthanthatofothermethods,butresults
of a comprehensive validation study show good agreement with literature data
[72]. The SGA method allows pKa measurement within 4 min leading to high
29
Chapter1
throughput capacity. The present available automated instrument (Sirius T3)
containinganautoloadermodule(roboticarm)utilizesfour48‐positionvialtrays
forsamples.Itenablesthemeasurementof240compounds/day.Lowsolubility
andlowmolarabsorptivitymaybelimitationsoftheSGAmethod.
1.3.1.4. Co‐solventmethod
DeterminationofpKausingtheabovediscussedmethodsisoftenhinderedbythe
lowwatersolubilityofthesamples.Itisafrequentproblemtodaysincethenew
moleculesindrugresearcharelesswater‐solubleandmorelipophilic.Forwater
insoluble compounds, the co‐solvent method can be used. In this approach, the
apparent ionization constants, psKa values, are measured in different ratios of
organic solvent/water mixtures. The aqueous pKa value is obtained by extra‐
polationtozeroorganiccontent.Theco‐solventmethodisprimarilyusedinpH‐
metry,butitcanbeappliedinUV‐spectroscopyandCEtechniquesaswell.
Manywatermiscibleorganicsolventshavebeenusedsuchasmethanol(MeOH),
ethanol (EtOH), propanol, DMSO, dimehtylformamide (DMF), acetone, and
tetrahydrofurane(THF).Sincemostliteraturedatahavebeenaccumulatedfora
MeOH/water solvent mixture and it is generally accepted that MeOH shows a
solvationeffectclosesttowater,MeOHisnormallychosenasanorganicsolvent
ofchoice[11,16,68].
Different extrapolation methods are known, but the Yasuda‐Shedlovky (YS)
extrapolation has proven to be the most reliable. Here, a linear correlation is
establishedinaplotofpsKa+log[H2O]vs.a/ε+b,wherelog[H2O]isthemolar
water concentration of the given solvent mixture, ε is the dielectric constant of
the mixture, and a and b are the slope and intercept, respectively. The aqueous
pKavaluescanbeobtainedforlog55.5and1/78.3,themolarconcentrationand
dielectric constant of pure water, correspondingly. The dielectric constant of
MeOH/watermixturesislowerthanthatofwaterandtheextentofionizationis
suppressed,thus pKa values of acids are shifted higher while those of bases are
toward lower values. The slope of the YS relationship is positive for acids and
negative for bases. The YS procedure offers many benefits over the traditional
plotofpsKavs.Rw(wt%oforganicsolvent)whichoftenshowsa“hockey‐stick”or
“bow”shape,sometimesatRw>60wt%anS‐shapecurve.Properelectrodecali‐
brationusingfourparameterproceduresinthesolventmixtureiscrucial[73].
Accordingtoacomprehensivevalidationstudy,thereproducibilityandprecision
of the method, based on 431 separate titrations in the interval of 15‐65 wt%
MeOHcontentusing25modelcompounds,wasfoundtobegood(SD=±0.05).
Extrapolationfromamethanol‐richregion(Rw:40‐60wt%)givesanerrorinpKa
notgreaterthan±0.2forweakacidsand±0.1forweakbases[74].
Sincenotallcompoundsdissolveinasingleorganicsolvent(e.g.methanol),anew
multicomponent co‐solvent system significantly improving the solubility of
pharmaceuticalcompoundswasrecentlydevelopedforpKadetermination[75,76].
The mixture consists of an equal volume of MeOH, dioxane, and acetonitrile
30
Physicochemicalprofilingindrugresearchanddevelopment
(referredtoasMDM)dilutedinwatertoobtaintherequiredco‐solventsystem.
This system enables pKa measurements by potentiometry (and also by UV/pH
titration) for a wide range of poorly soluble compounds. Since solubility
considerablyincreasesintheMDMsystem,measurementscanbeperformedina
lowerproportionoforganicsolvent,thusthelong‐distanceextrapolationcanbe
avoided.ThelinearityoftheYSrelationshipisvalidupto55wt%MDMcontent.
Validation based on 50 compounds showed good reproducibility
(SD=±0.01‐0.08)andtheagreementofpKavaluesextrapolatedbythismethod
withvaluesmeasuredbyothermethodsisverygood(<0.10unit).
The SGA method has been extended with measurements in 20 wt % MDM
content, and general calibration equations were set up for acids and bases
(pKa(aqueous)=apsKa (20%MDM)+b),soasinglepointestimationmayproviderapid
aqueous pKa values for water‐insoluble compounds in the early phase of drug
research[76].
1.3.1.5. Decisiontreeformethodselection
The selection of a suitable method must be based on the properties of the
compoundtested.Figure1.12showsasimpledecisiontreeformethodselection
usedinthelaboratoryoftheauthor[4].
compound
pH-metry
YES
UV/pH titration
0.5 mM solubility
in water
NO
pH sensitive
UV spectrum
Co-solvent method
pH-metry
YES 1-5 µM solubility
in water
NO
0.5 mM solubility
in MDM/water
YES
NO
Single point
estimation
80% DMSO
pH-metry
YES
YES
0.5 mM solubility
in 80% DMSO
NO
1-5 µM solubility
in MDM/water
NO
YES
Co-solvent method
UV/pH titration
NO
pKa prediction
Figure1.12.DecisiontreeformethodselectionofpKameasurement
31
Chapter1
Table1.4.MethodsforpKadetermination
Sample
Method
potentiometry
UV/pH
titration
solu‐
amount,
bility,
mg
mM
3‐5
Throughput
high
purity
speed1
capa‐ Precision
city2
>0.5 necessary 20‐30 10‐12
Instrumen‐
tation
high
GLpKa,SiriusT3
(Sirius,UK)
traditional
1‐2
>0.01 necessary
6‐8
hours
1
medium
pH‐meter+
spectrophotometer
automated
1‐2
>0.01 necessary
30
10‐12
high
GLpKa+D‐PAS,
SiriusT3(Sirius,UK)
NMR/pH
titration
1‐2
>0.5
2‐3
high
NMRspectrometer
singlechannel
<<1
>0.01
not
necessary
30
20
medium
CE
multiplexed
<<1
>0.01
not
necessary
30
150 acceptable
CePro9600
(CombiSep)
>0.01 necessary
4
240 acceptable
Profiler‐SGA,
SiriusT3(Sirius,UK)
CE*
SGA†
1
*CapillaryElectrophoresis
†SpectralGradientAnalysis
not
2‐3
necessary hours
1
min/comp.
sample/day
2
1.3.2. logSdetermination
Severalmethodshavebeendevelopedforthemeasurementofbothequilibrium
and kinetic solubility including traditional and high throughput techniques. Ex‐
cellent reviews [7,16,39,41] have surveyed the state‐of‐the‐art techniques.
Below,afterashortsummaryofkineticsolubilitymethods,approachesforequili‐
brium solubility measurement are discussed focusing on good laboratory
practice(GLP).
1.3.2.1. Methodsfordeterminationofkineticsolubility
Concerning the large number but small content, samples in the early phase of
drug discovery are subjected to compound‐saving and HT methods which are
suitable for the measurement of kinetic solubility. In the turbidimetric method
introduced by Lipinski et al. [77] small aliquots (0.5 μl) of DMSO stock solution
areaddedat1min.intervalstoaqueousbuffers(originally,2.5mlofpH7phos‐
phate buffer) until the compound precipitates from the solution reaching the
maximal (but not yet the equilibrium) solubility. The turbidity caused by the
precipitationismeasuredbylightscatteringinthe620‐820nmrangewithaUV
detector. In nephelometric [78], direct UV [79] and ultrafiltration‐LC/MS [80]
methodstheaboveprincipleisadaptedto96‐wellplateusingdifferentdetectors
(nephelometer,diodearrayUVand MS,respectively).Inthe twolater methods,
32
Physicochemicalprofilingindrugresearchanddevelopment
the precipitate is separated from the solution by filtration (or centrifugation)
beforetheconcentrationmeasurement.Theultrafiltration‐LC/MStechniquehas
theadvantageofhighsensitivityandthecapabilityofhandlingimpuresamples.
Commercially available instruments (Nephelostar, BMG; Nepheskan Ascent,
Thermo Labsystem, μSOL, pION) use fully automated liquid dispensing systems
and provide high capacity (measurement of 200‐300 compounds/day). The
presence of DMSO in the kinetic solubility experiments may considerably affect
theresultsinahighlycompound‐dependentway,thusitispracticaltokeepthe
DMSOataminimumlevel(lessthan0.5%).
Themaindisadvantagesofkineticsolubilitymeasurementsarethelackofstan‐
dardization,poorreproducibility,anddifficultiesinthecomparabilityofresults.
1.3.2.2. Methodsfordeterminationofequilibriumsolubility
1.3.2.2.1. Saturationshake‐flaskmethod(SSF)
The SSF method is the standard approach for the determination of equilibrium
solubility which when properly performed provides high quality data. It is a
simple but very time‐consuming procedure and requires lots of manual work.
The solution of the tested compound containing excess solid is prepared in
aqueous buffer using a small (2‐5 ml) glass vial. The heterogeneous system is
cappedand vigorouslystirredatachosentemperature(usually 25°Cor 37°C)
foraspecifiedtime(24,48horlonger)untiltheequilibriumhasbeenreached.
After that, the two phases (solid and liquid) are separated by sedimentation,
centrifugation, or filtration. Upon diluting sample aliquots with the solvent, if
necessary, the concentration of the saturated solution is measured by an
appropriate method, most frequently by UV spectroscopy or HPLC. Despite the
longevityofSSFuse,thereinnoacceptedstandardwaytocarryoutthismethod.
Published solubility studies show great differences in the experimental
conditionsused,particularlyconcerningthetimeofequilibration,themethodof
phaseseparation,andthecontrolofpHduringthemeasurement[37,38,41].
Recently, in a comprehensive study published by Baka et al. [81] the most
important experimental factors influencing the measured equilibrium solubility
bytheSSFwereinvestigated(seesomeresultsinSection1.4.2)andastandardized
protocolwasproposedforGLP[82].Thefollowingconditionsaresuggested:
 themeasurementmustbecarriedoutatcontrolledtemperaturewithprecision±0.1°C,
 SörensenphosphatebuffercanbeusedbetweenpH3‐7,whileBritton‐Robinsonbuffer
canbeusedinawiderpHrangefrom2.5to11.5.HClofappropriateconcentrationcan
beusedbelowpH2.5,
 thepHofthesolutionmustbecarefullycontrolledduringthemeasurement,advisably
beforeandaftertheequilibration,
 to avoid the difficulties in sampling, only a small (~ 5‐10 mg/5 ml) excess of solid
shouldbepresent,
 aminimumof24hisnecessarytoreachtheequilibrium,thistimeshouldconsistof6h
ofstirringand18hofsedimentation,butincaseofverysparinglysolublecompounds
longer stirring time may be necessary for equilibrium, so in the most rigorous
33
Chapter1
application of SSF, the required time of equilibration must be determined from
compoundtocompound,
 the safest technique of phase separation is sedimentation which assures a hetero‐
geneous system until equilibrium has been achieved; if an opalescent solution is
formed then the phase separation can be done by centrifugation while the most
erroneousfiltrationshouldbeavoided(seeresultsinSection1.4.2below),
 acompoundexistinginameta‐stablepolymorphformcanbetransformedintoamore
stableoneduringthedurationofsolubilitymeasurement,thustheanalysisofthesolid
phase(byX‐raypowderdiffractionorthermo‐gravimetricmethods)attheendofthe
experimentishighlyrecommended.
Using the above listed conditions the equilibrium solubility of more than 50 com‐
poundswasdeterminedwithastandarddeviationoflessthan4%inourlaboratory.
1.3.2.2.2. Potentiometricmethods
Theprincipleofthepotentiometricmethodsisbasedonthatcharacteristicshift
ofthetitrationcurvecausedbytheprecipitationoftheunionizedformofacom‐
pound from a solution. Potentiometric titration was introduced for equilibrium
solubility measurement by Avdeef et al. [83‐85]. The dissolution template tit‐
ration(DTT)methodusespKaandlogPvaluesasinputparameters.LogPisused
to estimate the intrinsic solubility based on a Hansch‐Yalkowsky type equation
(log So = 1.17 – 1.38 log P). Using the pKa and the estimated intrinsic solubility,
the DTT procedure simulates the entire titration curve before the assay begins.
Thiscurveservesasatitrationtemplate(theoptimalquantityofthetestedcom‐
poundforthetitrationissuggestedbythesimulation)andalsoasaguideforthe
righttitrationprotocol(howtheinstrumentdispensesthetitrantandcollectsthe
pHdata)inthecourseofthetitration.ThetitrationstartsatpHvalues,wherethe
compoundisunionizedandformsasuspension(solidmaterialispresentinthe
solution).Thetitrantisdispensedaccuratelyandslowlyintotheslurry,todrive
thepHofthesolutioninthedirectionofdissolution.Typically,a3‐10h(some‐
timeslonger)timeframeisrequiredfortheentireequilibriumsolubilitydatacol‐
lection(20‐50pHpoints)[41].Themethod,whenperformedwiththepSOLtit‐
rator(pION,US),providesaprecisesolubility‐pHprofilewithoutassumingaHH
relationship and is much faster than the SSF method but still has a very low
throughput.
The novel potentiometric procedure (CheqSol) has been developed recently for
rapid measurement of solubility using the instrument called the GLpKa‐D‐PAS
(Sirius, UK). In this method, the equilibrium solubility is actively sought by
changing the concentration of the neutral (unionized) form of a compound by
addingacidorbasetitrantsandmonitoringtherateofthechangeofpH,dueto
precipitation or dissolution in a process called “Chasing Equilibrium”. In this
method,thetitrationisstartedatpHvalue,wherethecompoundisfullyionized
and dissolved and performed toward the direction of pH where the unionized
form precipitates. The turbidity of the solution caused by the precipitation is
detectedwithafiberopticdipprobe.Withthismethodboththekineticsolubility
andtheequilibriumsolubilitycanbedetermined.Thekinetic solubilityvalueis
34
Physicochemicalprofilingindrugresearchanddevelopment
obtained from the concentration when the first precipitation of the unionized
form appears in the solution, while the equilibrium solubility is obtained from
actively seeking the equilibrium pH where an equal amount of the sample is
precipitating and dissolving per unit of time [86]. The CheqSol method is faster
(typically 30–60 min/compound) because the intrinsic solubility is determined
instead of the entire solubility‐pH profile. Then HH equation is used for the
calculationoftheapproximatelogS/pHprofile.ItwasvalidatedagainsttheSSF
methodandexcellentagreementofsolubilityresultswasfound[87].
1.3.2.2.3. μDISSmethod
A miniaturized rotating disk dissolution instrument, called μDISS ProfilerPLUS
(pION,US)hasbeendevelopedforcharacterizingtheintrinsicdissolutionratein
early preformulation. This apparatus is also suitable for the measurement of
equilibriumsolubilityofsparinglysolublecompounds,providedenoughmaterial
isusedtomaintainthesaturation[41,88,89].Inthisprocedure,5mgofdrugare
compressed into pellets and inserted into a rotating disk carrier containing an
embedded magnetic stir bar at its bottom. This assembly is placed into a glass
vialfilledwithasmallvolume(1‐3ml)ofaqueousbufferasthedissolutionmedi‐
um.TheconcentrationismeasuredwitharapidinsitufiberopticUV(diodear‐
ray) detector. The instrument employs six parallel dissolution vessels and eight
channels of UV detectors which provide better capacity above the SSF method.
Overthehighprecision,furtheradvantagesofthismethodare:(i)anypolymorph
changesduringdissolutioncanberecognizedand(ii)thelongerincubationtime
neededtoestablishthetrueequilibriumofthemoststableformofasolidmaybe
evidentinthedissolutioncurve[39].
1.3.2.2.4. Highthroughputmethods
SomemethodssuitableformediumorhighthroughputdeterminationoflogSwere
also described. The miniaturised shake‐flask (MSF) method developed by Glomme
etal. [90,91] is a compound saving and fast method, thus it is frequently used in
pharmaceutical companies. Typically, 0.1‐0.2 mg solid powder is introduced to a
speciallydesignedfilterchamberandasmall(e.g.2ml)volumeofaqueousbuffer
isadded. Purpose‐built filtercaps arefirmly attached andthe vials areshaken at
constant temperature for 24 h. The filter‐containing cap compartments are then
depressedtoeffectseparationofthesolidandthetopcompartmentsolutionsare
analyzed by fast gradient RP‐HPLC. The throughput is just medium, as 20
compounds/week can be measured. The MSF method was further developed for
HTmeasurementsbyZhouetal.[92]wherea96‐wellplateisusedasthesourceof
thesamplesandDMSOstockswereevaporatedviaaGeneVacevaporator.
Those96‐wellplatebasedHTmethods(originallydevelopedforkineticsolubility
measurement),wheretheincubationtimeislongenough(e.g.μSOLmethod,[39])
and the effect of DMSO content is eliminated, are also suitable for equilibrium
solubility determination. Generally, in these modified‐microplate methods the
24h incubation time is adequate to reach the solubility equilibrium [39].
35
Chapter1
However,duetosmallvolumes,theprecisepHcontrolduringthemeasurement
may be problematic. In the lyophilized solubility assay (LYSA) the sample is
dispensed into a microtiter plate along with 10 mM DMSO solution then the
organicsolventisremovedbylyophilizationandaqueousbufferisadded.During
a 24 h incubation period the plate is agitated by a shaking mechanism, then
filtratedandtheconcentrationismeasuredusingaUVplatereader[93].Another
promising HT procedure is the PASS (Partially Automated Solubility Screening)
method,wherethecompoundsaresuspendedinheptaneanddispensedintothe
platewells,thenheptaneisevaporatedbeforebufferisadded[94].
1.3.2.3. Specialapplications
Themethodsdescribedabovehavebeenappliedforspecialpurposes.Potentio‐
metric titration, according to the CheqSol approach, has been reported to study
the solubility of polymorphs. A new method named “potentiometric cycling for
polymorph creation” (PC)2, was developed to generate the most stable poly‐
morph in aqueous solution [95]. It was applied to sulindac producing two
polymorphs including a new, more stable one. It was found that their intrinsic
solubilitydifferbyafactorofseven,whichismuchlargerthanthatofanyearlier
measureddifferencebetweenpolymorphs.
Table1.5.MethodsforlogSdetermination
Method
Detection
LOD,
μg/ml
Throughput
speed,
capacity, Precision
min/comp. samp./day
Instrumen‐
tation
forkineticsolubility
turbidimetric
UV
nephelometric
lasernephe‐
lometer
5
15
5
4
direct‐UV
UV
2·10‐3
ultafiltration‐
LC/MS
MS
0.1
50
low
300
low
Nephelostar,
Nepheloscan
4
300
medium
μSOL
6
200
medium
LC/MS
forequilibriumsolubility
SSF
UV;HPLC
Potentiometric
DTT
pH‐metry
36hours
<1
high
‐3
5·10 3‐10hours
1‐5
high
pSOL
CheqSol
pH‐metry
0.1
30‐60
10‐15
high
GLpKa;SiriusT3
UV
1
24hours
6
high
μDISS
ProfilerPLUS
MSF
UV;LC/MS
1
3‐100
medium
μSOL
UV
0.1
24hours
18‐24
hours
100
acceptable
μSOL
μDISS
modified‐plate
HT
1
LYSA
UV
1
24hours
100
acceptable
PASS
UPLC
103
24hours
100
acceptable
36
Physicochemicalprofilingindrugresearchanddevelopment
Thebiorelevantsolubilityvaluesaremoreandmorerequiredindrugdiscovery
and development (DD&D). An optimized 96‐well HT UV solubility method was
adaptedtomeasuresolubilityofdrugsinbiorelevantmediasuchasFaSSIFand
FeSSIF solutions [96]. The method provides reliable data using a very small
amount of sample and small volumes of the expensive FaSSIF/FeSSIF compo‐
nents.TheμDISSmethodwasalsofoundusefulformeasurementinbiorelevant
media and temperature [97]. The study has revealed that the majority of the
testeddrugsexhibitedhighersolubilityinthesemediathaninpurebuffers.
1.3.3. logPdetermination
Since log P is the oldest parameter in physicochemical profiling, several well‐
established experimental methods are available for its determination. Vast
amountsofliteraturehavedescribedthetheoryandpracticeoftheusedmethods
[e.g.11,46,98‐101].AspectsfromtheGLPguideforlogPmeasurementshavealso
been published [98,101,102]. The recent reviews provide a comprehensive
surveyaboutthelatestdevelopmentsinHTtechniques[6,9,14‐16].
TwotypesofmethodscanbedistinguishedforlogPdetermination:(i)thedirect
approaches, where log P is directly obtained from the measured data (shake‐
‐flask, stir‐flask, filter chamber, dual‐phase potentiometric, etc.) and (ii) indirect
(chromatographic, CE) techniques, where the measured parameter has a linear
relationshipwithlogPandlogPiscalculatedusingcalibrationequations.Inthis
chapter,thedirectmethodsareoverviewed,outoftheindirectmethodsonlyTLC
ispresented,whileotherslikeHPLC,MECK,etc.usedforlogPmeasurementare
discussedelsewhereinthebook.
Inordertofacilitatethecomparisonoftheircapacity,Table1.6summarizesboth
typesofmethods.
1.3.3.1. Shake‐flask(SF)method
The traditional SF method is the reference and most widely used approach of
logPdetermination.
Procedure.Inadvance,thetwophases(n‐octanolorotherusefulpartitionorganic
solvent that is immiscible with water and aqueous buffer) must be mutually
saturated with vigorous agitation then filtered or centrifuged. The tested
substanceisdissolvedintheaqueousphaseandintroducedintoanappropriate
glassvial.Octanol(orotherorganic solvent)isaddedinarequiredvolumeand
the system is shaken at a constant temperature for a period long enough for
equilibrium to be achieved (generally 1 h). After separation of the phases by
centrifugation, the concentration is measured using an appropriate method,
mostlyUVspectroscopy.Concerningthedifficultiesofthepreciseanalyticalwork
with octanol, it is a common practice to measure the concentration decrease in
theaqueousphasebydetectingtheabsorbancebeforeandafterthepartition.
Accuracy, sources of the experimental error. The SF method is suitable for logP
measurementintherangefrom–2to5havingaSD<0.05,providedthatoptimal
37
Chapter1
experimentalconditionsaremaintained.Manyfactorscanaffectthereliabilityof
the measured log P values increasing the experimental error. One of them is
undoubtedly the applied extreme phase ratio necessary in the case of lipophilic
compounds(logP>3).Accordingtoourexperiences,thehighestphaseratiothat
can be used without a considerable increase in error is R = 500 (e.g. 50ml
aqueousbuffer:0.1mloctanol).However,intheoppositecasewithhydrophilic
compounds, when more octanol has to be used, the sampling from the lower
aqueous phase may be problematic, thus it is advisable to remove the upper
octanollayerbeforealiquotsaretaken.Glassandsurfaceadsorption,formation
of stable emulsions, and the presence of impurities in the sample have often
influencedtheresults.
Advantages/drawbacks. The main advantage of the SF method is its simplicity,
sufficient accuracy, and applicability to non‐ionizable compounds. But it has
some well‐known shortcomings, such as being tedious and time‐consuming,
difficultieswithmaintainingaconstanttemperatureduringthewholeprocedure,
requiringrelativelyhighamountsofsampleandsolvent.TheSFmethodcannot
be used for UV inactive compounds unless alternative detection methods are
employed,andsoon.
1.3.3.2. Potentiometricmethod
Dualphasepotentiometrictitrationusingautomatedinstrumentshasbecomethe
“goldstandard”oflogPdetermination(forionizablecompounds)inthepastten
years [11,60,100]. It consists of two titrations of the tested compound. One is
performedwithoutthepartitionsolventandprovidestheaqueouspKavalue.The
second is done using the same conditions but in the presence of a partition
solvent(e.g.octanol)withintensivestirringupontitrantaddition,whilestopping
it when the pH is measured. If the unionized form of the compound partitions
intooctanolthenthetwotitrationcurveswillbedifferent,duetoashift(similar
to what was discussed in the co‐solvent pKa method). From the dual‐phase
titration, the apparent poKa value is obtained. Log P is calculated from the
differences in pKa values and the phase ratio. A large shift indicates high
lipophilicity(seealsoinSection1.4.3).
This method allows the log P determination in a range from ‐2 to 6, with very
high precision (SD = ± 0.01). The agreement with the SF method is excellent
according to validation studies [103,104]. However, it has limited capacity and
cannot be used for compounds with pKa out of the established measurable pH
range.Afurtherdrawbackisthatonlyalimitedphaseratiocanbeapplied(inour
practicewithGLpKa:20mlwater:0.05mloctanolforlipophiliccompoundsand
5mlofwater:15mlofoctanolforhydrophiliccompounds).Anewlydeveloped
instrument(SiriusT3)hasfurtherincreasedefficacyandmeasurementispossible
inaslowas1mlaqueousphase.
38
Physicochemicalprofilingindrugresearchanddevelopment
1.3.3.3. IndirectlogPmeasurementmethods
Because of the drawbacks and limitations of direct methods, numerous
alternative procedures have been developed and applied. Several micellar,
microemulsion, vesicle electrokinetic chromatographic systems, and reversed‐
phase chromatographic methods (RP‐TLC, RP‐HPLC) can be used to estimate
lipophilicity. Some excellent reviews on the use of separation methods for
indirectlogPdeterminationhavebeenpublished[e.g.105‐106].
AlthoughRP‐HPLCismorewidelyusedtechniqueforlogPestimation[105],RP‐
TLC undoubtedly has some unique advantages, including use of less expensive
laboratoryequipmentandbeingeasytoperform.Simultaneousrunningof15‐20
compounds on one plate can significantly reduce the analysis time per com‐
pound. The method is based on the linear relationship between logP measured
bytheSFmethodandthelogarithmofchromatographicretentionexpressedas
the RM value. RP‐TLC has been successfully applied for logP measurement of
highlylipophilicmoleculesusingcalibrationequationsobtainedwithstructurally
related compounds [e.g. 107‐109]. Recently, a validated RP‐TLC method was
proposedforparallelestimationofthelipophilicityofchemicallydiverseneutral
compoundsorweakacidsandbases[110].Tocoverawiderangeoflipophilicity,
two optimized RP‐TLC systems were used: one for moderate lipophilic com‐
pounds(logP=0‐3)andanotherforhighlylipophilicmolecules(logP=3‐6).
Two chemically diverse sets of compounds were selected to set up the general
calibration equations. The method was tested with 20 randomly selected drugs
and good agreement with SF results were found (SD < 0.15). With automated
sampling and imaging detection of the compounds the method can be regarded
as a possible alternative for rapid and acceptable accurate estimation of lipo‐
philicityofdrugcandidatesintheearlyphaseofDD&D.
1.3.3.4. Highthroughputmethods
Attempts have been done to miniaturize the traditional SF method into a
microtiter plate platform with robotic liquid handling and HPLC/UV [111] or
LC/MS [112] detection. In these techniques the partitioning process is
transferred to 96‐deep well plates and after equilibration, the detector signal
produced by a sample from the octanol phase is divided by the signal from the
aqueousphase.InareviewbyKerns[7],thecriticalelementsofthesemethods
arediscussed.AcommerciallyavailableautomatedplatebasedinstrumentforHT
logPdeterminationiscalledtheAlogP(Analiza,US).
Regarding indirect (HPLC and MECK) log P methods, additional successful
strategies were applied to increase the throughput and speed up the determi‐
nation time [16]. For example, the use of short columns and a high flow rate in
HPLC,usageofUPLC,andmultiplexedMECKhavebeenreported.Thesemethods
aresurveyedintheChapter2ofthisbook.
39
Chapter1
1.3.3.5. Decisiontreeformethodselection
Themethodselectionstrategyfollowedinthelaboratoryoftheauthorisshown
inFigure1.13.
compound
pH-metric
YES
ionizable
YES
solubility >1mM
NO
NO
YES
chromophore
expected log P
-2
5
YES
Shake-flask
NO
NO
RP-TLC
RP-HPLC
log P prediction
Figure1.13.DecisiontreeformethodselectionoflogPmeasurement
Table1.6.MethodsforlogPdetermination
Method
logP
range
Throughput
sample
amount,
speed,
capacity, Precision Instrumentation
mg
min/comp. samp./day
direct methods
shake-flask
traditional
automated
(96-well plate
platform)
potentiometric
-2 ↔ 5
2-10
180-360
2
high
-2 ↔ 5
1-5
10
100
acceptable
AlogP
(Analiza Inc.)
-2 ↔ 6
1-5
60
20
high
GLpKa, SiriusT3
indirect methods
40
RP-TLC
0↔6
1-3
120
50
medium
RP-HPLC
-1 ↔ 6
0.01
15
100
acceptable
MEEKC
-1 ↔ 7
<< 1
15
150
acceptable
CePro 9600,
MCE 2000
Physicochemicalprofilingindrugresearchanddevelopment
1.4.
CASESTUDIES
Inthischapterwepresentsomeusefulexamplesofphysicochemicalprofilingfor
the illustration of methods discussed above. Mainly, such problematic compo‐
undshavebeenselectedwheretheroutinemeasurementisdifficultorhindered
bycertainreasons.ThemostfrequentdifficultiesinthepKa,logS,andlogPde‐
terminationarethelowsolubility,instability,lackofUVactivityandpolymorph
transition of the compound. The case studies introduced below can provide a
possibletemplateforthemeasurementof“difficulttomeasure”molecules.
1.4.1. pKadetermination
Case1.
Sample:
Method:
RG‐1503
co‐solventmethod;potentiometrictitrationin
methanol/watersystem
Instrumentation:GLpKa(Sirius,UK)
Thecompoundisamultiproticmoleculecontainingfourionizablegroups(Figure
1.14a:A‐D).ThetwopiperazineNatomsandthepyridineNareprotonaccepting
basic centers, while the arylsulfonamide moiety represents a proton releasing
acidicgroup.Theaqueoussolubilityofthecompoundislessthan0.5mMandit
has no useful pH‐dependent UV spectrum in the pH range 3‐10, thus neither
potentiometrynorUV/pHtitrationinanaqueousmediumcanbeused.
Figure1.14.Co‐solventpH‐metryforpKadetermination:
(a)structureofthesample(RG‐1503),(b)titrationcurvesindifferent
methanol/watermixtures,(c)Bjerrumplots,(d)YSextrapolationcurves
41
Chapter1
The pKa values were measured by potentiometry using the co‐solvent method
(seeSection1.3.1.4).Sixtitrationswereperformedinmethanol/watermixtures
(40‐60 wt%) between pH 1.5 ‐ 12.5 in ~ 1 mM concentration of the sample, at
0.15 M (KCl) ionic strength, at 25.0 ± 0.1 °C temperature, under N2 atmosphere
(Figure 1.14.b). From the obtained psKa values, the aqueous pKa values were
calculatedbyYSextrapolation(Figure1.14d).
Results:pKa1=2.02±0.22(Bgroup);pKa2=3.03±0.09(Cgroup);
pKa3=7.35±0.03(Agroup);pKa4=11.40±0.09(Dgroup).
Case2.
Sample:
nitrofurantoin
Method:
co‐solventmethod;UV/pHtitrationinMDM/watersystem
Instrumentation: GLpKa+D‐PAS(Sirius,UK)
Nitrofurantoinisawater‐insolublecompoundwhichhasoneacidicgroupandex‐
hibits a pH‐dependent UV spectrum. The pKa value was measured in an
MDM/watersystembecauseitssolubilityishighenoughinthissolventmixture
for the spectroscopic determination. A stock solution was prepared in 10 mM
concentrationwithMDM,50μlofthisstocksolutionwasusedforthetitrationin
15mlof20‐50wt%MDM/watermixturesbetweenpH3‐10,at0.15M(KCl)ionic
strength,at25.0±0.1°Ctemperature,underN2atmosphere.Figure1.15.cshows
thepsKavaluesusedforYSextrapolation.TheextrapolatedaqueouspKa valueis:
6.87 ± 0.01 (R2= 0.9958), which is in good agreement with literature data
measuredbyothermethods[76].
(a) nitrofurantoin
(b)
O
N
N
0.8
0,8
O
0.6
0,6
O
Absorbance
O 2N
1.0
1,0
H
N
0.4
0,4
Absorbance
0.2
0,2
0.0
0,0
250
(c)
300
350
400
450
500
Wavelength (nm)
(d)
17.6
25.4
36.6
43.4
48.8
9.1
psKa± SD
7.05
7.16
7.33
7.46
7.58
± 0.05
± 0.05
± 0.05
± 0.05
± 0.05
9.0
psKa + log[H2O]
t%)
R(w
psKa + log[H2O] = 60.4/ε
60.4/? + 7.837
8.9
8.8
8.7
8.6
13
14
15
16
1/ x 1000
Figure1.15.Co‐solventUV/pHtitrationforpKadetermination:
(a)structureofthesample(nitrofurantoin),(b)pH‐dependentUVspectra,
(c)apparentpKavaluesindifferentMDM/watermixtures,(d)YSextrapolation
42
Physicochemicalprofilingindrugresearchanddevelopment
Case3.
Sample:
Method:
lisinopril
potentiometrictitrationinaqueoussolutionandin
methanol/watersystem;NMR/pHtitration
Instrumentation: GLpKa+D‐PAS(Sirius,UK);VarianInova600MHz
spectrometer(PaloAlto,CA)
Lisinoprilisatetraproticcompoundhavingtwoacidic(carboxyl)andtwobasic
(a primary and a secondary amine) groups. Figure 1.16a shows the ionization
processesofthemolecule.Thedissociationofthetwocarboxylgroupsishighly
overlapping.Thesolubilityoflisinopril(0.22M)allowsthedeterminationofpKa
valuesbythestandardpotentiometricmethodinaqueousmedium.However,the
first dissociation constant falls into the low pH range which may cause
uncertaintyofthemeasurement.
For the characterization of the acid/base property of the molecule, three
independentmethodswereapplied:potentiometryinaqueoussolutionandina
methanol/watersystem,aswellasNMR/pHtitration.Inaqueousmedium,three
titrationswerecarriedoutina2mMconcentrationsolution,betweenpH1.8‐12,
at 0.15 M (KCl) ionic strength, at 25.0 ± 0.1 °C temperatures, under N2
atmosphere.SincepKa1valuefallsbelowthelowerapplicabilitylimit(<2)ofpH‐
metrictitration,thepKavalueswerealsomeasuredusingtheco‐solventmethod.
(a)
+ H+
COO
N
H
Lis 2-
N H3
+ H+
COO
N
O
NH3
NH 2
N
H
COO
HLis -
COO
N
N
H2
COO
O
H 2Lis
N H3
COO
N
H2
N
O
+ H+
COO
COOH
N
H2
C OOH
H 3Lis +
(b)
O
NH3
+ H+
N
N
O
COOH
H 4Lis 2+
HLis -
H 2Lis
100
Lis 2-
H 4Lis 2+
80
H 3Lis +
% Species
60
% Species
40
20
0
2
4
6
8
10
12
pH
Figure1.16.Protonationschemeof(a)tetraproticlisinopril,
(b)distributionofmacrospecies
43
Chapter1
The apparent pKa values of the COOH groups obtained in 14‐44 wt% metha‐
nol/water mixtures shifted up to the established measurable pH range, and a
reliableaqueouspKa1valuecouldbeobtainedbyYSextrapolation.
Fortheexactprotonspeciationoflisinopril,1HNMR/pHtitrationswithinsitupH
measurementswerecarriedout,usingthemostsimilarexperimentalconditions
aspossibleinpotentiometry.Thismethodwasusefultoassigntheconstantsto
thefunctionalgroups:pKa1andpKa2belongtotheCOOHgroups,pKa3referstothe
secondary amine (–NH–), and pKa4 shows the basicity of the primary amine
(–NH2)function.
The highly precise pKa values were calculated as an average of the best two
valuesobtainedbyindependentmethods(Table1.7).Thesevalueswereusedto
calculate the distribution curve of different protonated species of lisinopril
againstthepH(Figure1.16b).
Table1.7.ThepKavaluesoflisinoprilmeasuredbydifferentmethods
method
ionizationconstants
pKa1±SD
pKa2±SD
pKa3±SD
pKa4±SD
potentiometry
1.54±0.05
3.10±0.01
7.14±0.01
10.74±0.01
potentiometryinsolventmixtures
1.62±0.01
3.21±0.02
7.22±0.03
10.75±0.01
NMR/pHtitration
1.63±0.01
3.15±0.01
7.12±0.01
10.53±0.03
averageofthebesttwovalues
1.63±0.01
3.13±0.01
7.13±0.01
10.75±0.01
1.4.2. logSdetermination
Case4.
Sample:
hydrochlorothiazide
Method:
SSF
Instrumentation:RadiometerPH220pHmeter;LAUDAM20Sthermostat;
HeidolphMR1000magneticstirrer;JASCOV‐550UV/VIS
spectrophotometer
HydrochlorothiazideisabivalentacidwithpKavalues:8.75and9.88.Itsintrinsic
solubility(So)valuewasmeasuredatpH6.0usingtheSSFmethod[81].
First the So of the sample was measured according to a standard (literature)
protocol with the following conditions. Buffer: Britton‐Robinson (BR); solid
excess:smallamount;temperature:25.0±0.1°C;equilibrationtime:48hstirring
plus 24 h sedimentation; phase separation technique: sedimentation; concen‐
tration measurement: UV spectroscopy (λ=271 nm, A1%1cm: 696); number of
parallels:6.Result:So=556±13.2μg/ml.
Next,differentparametersofthisprotocolwereexamined,alwaysoneofthesix
parameters(bufferchoice,amountofsolidexcess,temperature,timeofstirring,
time of sedimentation, phase separation technique) was varied while the other
conditionswerekeptunchanged.
44
Physicochemicalprofilingindrugresearchanddevelopment
Effect of buffer solution.Three buffer solutions were used at pH 6.0. The results
areshowninFigure1.17a.ThestatisticalanalysishasindicatedthatSovaluesin
BR and Sörensen phosphate (I) buffers are in accordance, but the solubility in
Sörensencitrate(II)bufferdeviatessignificantly.Theionicstrengthofthislater
bufferisfourtimeshigherthanthatofBRorSörensenI.
Effect of phase separation. Alternative techniques to sedimentation such as cen‐
trifugation and filtration were studied. 12 samples were centrifuged after 48h
stirringat2000rpmfor10min,while12sampleswerefilteredthrough0.45μm
membrane filters. Results shown in Figure 1.17b are significantly different. The
highestdeviationiscausedbyfiltration.
Effectofequilibrationtime.Figure1.17cshowstheexperimentalresultsobtained
when(i)stirringtimewaschangedfrom30minto48hfollowedbya24hsedi‐
mentation;and(ii)sedimentationtimewaschangedfrom1hto24h,keepingthe
stirringtimeconstant(48h).Fromtheresultsitcanbeconcludedthatthetimeof
sedimentation plays a greater role in the development of equilibrium than the
timeofintensiveagitation.
Effect of temperature. The solubility of hydrochlorothiazide increases with the
temperature (Figure 1.17d). It is almost double at 37 °C than at 25 °C, which
underlines the need for solubility determination at biomimetic temperature as
well.
(a)
(b)
779
g / ml
600
580

560 ,
Solubility
540 516
520
500
480
460
0.5
Solubility,g / ml
565
Sörensen I.
564
508
1
563
661
591
556
sedimentation
centrifugation
12
24
48
Time [h]
Time,
h
g / ml
640
631 610
610 631
610
620 605 606

600 ,
Solubility
575 580
580
556
560
540
520
500
1
2
4
6
8 12 18 24
Solubility,g / ml
Time,
h
Time [h]
filtration
556
1200
6
(d)
580
523
2
680
660
640
620
600
580
560
540
520
500
Sörensen II.
1000
g / ml
800 ,
Solubility

600
Solubility,g / ml
(c)
Solubility,g / ml
Solubility,g / ml
900
800
700
g / ml
556
600
Solubility ,

500
400
300
200
100
0
Britton-Robinson
1036
556
450
400
200
0
15 °C
25 °C
37 °C
Figure1.17.Effectofexperimentalconditionsontheintrinsicequilibriumsolubilityof
hydrochlorothiazide:(a)buffersolution,(b)phaseseparationtechnique,
(c)stirringtime(uppergraph)sedimentationtime(lowergraph),(d)temperature
45
Chapter1
Case5.
Sample:
papaverinehydrochloride
Method:
SSF
Instrumentation: RadiometerPH220pHmeter;LAUDAM20Sthermostat;
HeidolphMR1000magneticstirrer;JASCOV‐550UV/VIS
spectrophotometer
Thesolubility‐pHprofileofpapaverinehydrochloridewasdeterminedinawide
pHrangeusinganew(shorter)protocolderivedfromtheSSFmethod:BRbuffer,
smallsolidexcess,25.0±0.1°C,6hstirringand18hsedimentation[42].
First,theintrinsicsolubilityofthesamplewasmeasuredathighpH(11.71)and
found 17 μg/ml (log So = 1.70 [log μM]). Then the equilibrium solubility (SpH)
valuesattwelvedifferentpHvaluesbetween0.06and8.02weredetermined.
FromlogSoandpKa(6.36)valuesthetheoreticallogSpH/pHprofilwasgenerated
by the HH equation. Figure 1.18 shows the excellent agreement between the
experimentaldatapointsandthepredictedHHcurve.Thisshapeistypicalfora
monovalentbase,wherethesolubilityincreaseswithadecreaseofthepH,asthe
freebasestartstoconverttotheprotonatedform.AtaroundpH3,thesolubility
of the papaverine hydrochloride salt reaches the maximum (pHmax), which is
limitedbythesolubilityproduct.BetweenpH2and3thereisaconstantvaluefor
the salt solubility. Below pH 2 the solubility of the salt decreases due to the
commonioneffect,causedbychlorideionsfromHClusedtoadjustthepH.
Figure1.18.Solubility‐pHprofileofpapaverinehydrochloride
46
Physicochemicalprofilingindrugresearchanddevelopment
This example proves that the HH equation can be used for the calculation of
solubility at physiological important pH values once the intrinsic solubility and
thepKavaluehavebeenpreciouslydetermined[42].
Case6.
Sample:
telmisartan
Method:
SSF
Instrumentation: RadiometerPH220pHmeter;LAUDAthermostat;Heidolph
MR1000magneticstirrer;JASCOV‐550UV/VIS
spectrophotometer
Thesolubilityoftelmisartanwasmeasuredindistilledwater(ordinarilypH~6)
andat37±0.1°Ctemperature(oneoftheconditionswheresolubilityisrequired
bytheregistrationauthorities).Solidmaterialat0.01gwasaddedto20mlfresh‐
lyboiledandcooledwaterandthenthenew(shorter)protocolderivedfromthe
SSFmethodwasfollowed.Aliquotsweretakenoutfromthesupernatantandthe
absorbance was measured without dilution at λ = 295 nm, in a cell with a 5cm
pathlength.TheconcentrationwascalculatedusingA1%1cm=510measuredsepa‐
rately prior to solubility measurement. From three parallel experiments, the
solubility of telmisartan was found as low as SpH=0.50±0.09μg/ml. The relati‐
velyhigherror(SD=±18%)isduetotheverylowsolubility(thelowestvalue
wecouldevermeasurebytheSSFmethod)andtheoccasionallyformedsupersa‐
turated solution, from which small (invisible) particles precipitated in the cell
uponabsorbancemeasurement.
Case7.
Sample:
maprotiline
Method:
SSFandCheqSol
Instrumentation: RadiometerPH220pHmeter;LAUDAthermostat;Heidolph
MR1000magneticstirrer;JASCOV‐550UV/VIS
spectrophotometerandGLpKa+D‐PAS
The precise intrinsic solubility of maprotiline base (pKa= 10.33) ‐ another very
sparingly soluble compound ‐ could not be determined by the SSF method. The
resultobtainedfromthreeseparate measurementsinBR bufferatpH11.5was
So=8.05±3μg/ml. The reason for the extremely high experimental error
(SD=±37%) is that a colloid, slightly opalescent solution (perhaps due to re‐
crystallization or supersaturation) was formed upon equilibration. This opale‐
scence could be eliminated by neither filtration nor centrifugation. So, the SSF
solubility result must be considered as an approximate value. Thus, the poten‐
tiometricmethod,namelytheChasingEquilibriumSolubility(CheqSol)wasalso
applied.Maprotilinewasadded(2mg)to10mlof0.15MKClsolutionthenpre‐
acidified with 0.5 M HCl to pH 2 where the compound was fully dissolved. This
solution was titrated with 0.5 M KOH until the solution became cloudy, which
indicatedtheprecipitationofthefreebaseform.Theoccurrenceofprecipitation
was detected using a spectroscopic dip probe then the solution was quickly
47
Chapter1
brought close to equilibrium by adding very small amounts of acidic or basic
titrants alternatively resulting in an oscillation between supersaturation and
subsaturation. The Bjerrum plot of titration is shown in Figure 1.19. While the
sample is fully dissolved, the experimental data fit well to the nonprecipitation
theoreticalcurve(a).Afterprecipitation,thepointslieclosetotheprecipitation
theoretical curve (b). The precipitation point is used to calculate the kinetic
solubilityvalue.Theintrinsicequilibriumsolubilitywasdeterminedfrom40data
pointswith8zeropHgradientcrossings.
The intrinsic solubility was obtained as average of 6 separate titrations,
So=5.8±0.3μg/ml.ThelowSDindicatesthehigherprecisionofthedataandthe
advantageoftheCheqSolmethodinthiscase.
Figure1.19.BjerrumplotofsolubilitydeterminationofmaprotilinebyCheqSolmethod.
(a)nonprecipitationtheoreticalcurve,(b)precipitationtheoreticalcurve
Case8.
Sample:
venlafaxineHCl
Method:
SSF
Instrumentation: RadiometerPH220pHmeter;LAUDAM20Sthermostat;
HeidolphMR1000magneticstirrer;JASCOV‐550UV/VIS
spectrophotometer
Venlafaxineisamonovalentbase(pKa=9.6),itshydrochloridesaltcanformdif‐
ferent polymorphs. The solubility of two polymorph forms (I and II) was inves‐
tigatedatthreepHvalues:4.9(unadjustedpHindistilledwater),8.9(BRbuffer),
and12(0.001MNaOH)at37±0.1°CtemperatureusingtheSSFmethod.There‐
sults are summarized in Table 1.8. The salt solubility is higher than 50 %
(g/100ml)inthecaseofbothpolymorphs.Theintrinsicsolubilityofvenlafaxine
48
Physicochemicalprofilingindrugresearchanddevelopment
measured at pH 12 was also found to be the same for both I and II forms
(So=460±10μg/ml).
Thediffractionanalysisofthesolidphasefilteredoutattheendofthesolubility
measurement revealed that polymorphs I and II equally converted to the same
crystal form of free base venlafaxine. This experience underscores the need for
analysisofthesolidphaseaftertheequilibriumstatehasbeenreached.
Table1.8.Solubility(g/100ml)oftwopolymorphformsofvenlafaxine
hydrochlorideatthreepHvaluesandat37°Ctemperature
venlafaxinehydrochloride
SpH(pH4.9)
SpH(pH8.9)
So(pH12.0)
FormI
>50
0.180
0.046
FormII
>50
0.208
0.046
1.4.3. logPdetermination
Case9.
Sample:
chlorpromazine
Method:
SSF
Instrumentation: LAUDAM20Sthermostat;Hawlett‐Packard8452AUV/VIS
spectro‐photometer
Chlorpromazine is a very lipophilic monovalent base (pKa: 9.24). The true logP
valuecannotbemeasuredathighpHvalues(>11.5)directlybytheSFmethod
becauseofthelowsolubilityofthefreebaseformofthecompoundathighpH.In
suchcases(whichistypicalamongdrugs),thelogDpHismeasuredatdifferentpH
valuesatwhichthemoleculepartiallyionizesanddissolvesbetterandthenitis
converted to the true log P using Equation 1.30b. The log DpH values of
chlorpromazineweremeasuredatthreepHvalues(7.4,8.0and8.5)inBRbuffer
astheaqueousphase,usinganR=200and100phaseratios(50mlbuffer:0.25
mloctanoland25mlbuffer:0.25mloctanol).Wefollowedthestandardprotocol
oftheSFmethod:1hintensiveshakinginashakingthermostat;phaseseparation
bycentrifugation(730gfor10min).Theabsorbanceoftheaqueousphasebefore
(Ao) and after (A1) the partition was measured by spectroscopy at λ = 254 nm.
The apparent partition coefficient is calculated according to DpH = [(Ao‐A1)/A1]R
[104].Thelipopilicity‐pHprofileisshowninFigure1.20.
Result:logP=5.13±0.10(n=18)
49
Chapter1
Figure1.20.Lipophilicity‐pHprofileofchlorpromazine
(pointsrepresenttheexperimentallymeasuredlogDpHvalues)
Case10.
Sample:
deramciclane
Method:
potentiometricmethod
Instrumentation: PCA101(Sirius,UK)
Deramciclane was an original anxiolytic molecule developed by EGIS (Hungary)
inthelate‘90s,whichunfortunatelyfailedfromclinicalphaseIII.Itisasparingly
solublemonovalentbase(pKa=9.61).IthasaveryweakUVabsorption(lowspe‐
cificabsorptivity)thuslipophilicitydeterminationbytheSFmethod(asdonein
Case9)ishindered.ThelogPvaluewasmeasuredbydual‐phasepotentiometric
titrationat25.0± 0.1°C temperature,underN2atmosphere.Sixtitrations were
performedbetweenpH3and12,ina1mMconcentrationsolutionofthesample
using 15 ml water and 0.05 ml octanol phase (Figure 1.21a). From these
titrations the apparent pKa values (measured in the presence of octanol, poKa)
wereobtained.TheBjerrumplot(Figure1.21b)showsbigshifttowardlowerpH
values(typicalforbases)whichindicateshighlipophilicityofthesample.Thelog
P value is calculated according to the equation: log P = (10(pKa ‐ poKa) – 1)/r. The
extreme(octanol/water)phaseratio(r=0.0033)usedhereallowedlipophilicity
measurementashighaslogP=5.90±0.02(n=6).Accordingtoourexperiences,
this represents the upper limit of the pH‐metric log P determination method
[113].
50
Physicochemicalprofilingindrugresearchanddevelopment
Figure1.21.pH‐metriclogPdeterminationofderamciclane:
(a)titrationcurvesinthepresenceofdifferentamountsofoctanol,(b)Bjerrumplot
Case11.
Sample:
prostaglandinE1‐ethylester(PGEE)
Method:
RP‐TLC
Instrumentation: RP‐diC1silanizedplates,Merck#5747;Camag
microsampler;ShimadzuCS‐9301CPdensitometer
PGEE is an example for molecules where classical, standard methods cannot be
applied. Due to the lack of useful UV absorption (above λ > 230 nm) or an
ionizablegroup,neithertheSFnorpH‐metrycanbeused.LogPwasdetermined
byavalidatedRP‐TLCmethod.
O
COOC2H5
CH3
HO
OH
Measurementwasperformedon20cmx20cmplatesprecoatedwithsilanized
silicagelGF254asthestationaryphaseandmethanol/water(55:45)asthemobile
phase. Before use, the plates were washed with methanol (ascending deve‐
lopment),thendriedandheatedat160°Cfor1h.Thesamples(PGEEandcali‐
bration set) were dissolved in a 1 : 1 methanol/chloroform mixture (2mg/ml)
and 2 μl was spotted on the plate. The chamber was saturated with the mobile
phase for 30 min before use. After development the plates were dried and
evaluatedbydensitometry.
The calibration curve was set up using seven compounds [114] and obtained
fromthreeparallelruns:logP=3.508RM+0.968(r=0.995,n=21).ThelogPof
PGEEwascalculatedwiththehelpofthisequation.
Result:logPTLC=4.02±0.05(n=3).
51
Chapter1
1.5.
OUTLOOK
Concerning the role of physicochemical profiling in the future, we can certainly
predict that it remains an integrated part of drug research providing a simple,
cheap,andfasttoolfortheestimationofADMETparametersintheearlystageof
DD&D.
A higher level of automation (e.g. integration of several robotic platforms) and
highersensitivityofdetectionmethodscanbeexpectedleadingtotheincreaseof
the HT feature of the applied methods, but it must be synchronous with the
improvement of the reliability of the data determined. Next to this, the cost‐
effectivenesswillbethecriticalfactorintheselectionbetweenmethodshaving
thesamecapacity.
We can anticipate the increasing application of biorelevant experimental
conditions in physicochemical profiling. Standardization and validation of these
biomimeticsystemsareobviouslynecessaryinthenearfuture.
The use of in silico methods will be growing if further development of
computational approaches results in even more reliable data. For the in silico
methodsbasedonbigdatabasesthe qualityoftheinputofexperimentalvalues
mustbefurtherimproved.
Finally,moreeffectiveusageofphysicochemicalprofilingindrugresearchcanbe
promoted by including informative courses or seminars, for example, in higher
educationtostrengthenthisspecialfieldofmedicinalchemistryinacademia.
Acknowledgement
IwouldliketothankmycolleagueGergelyVölgyi,PhDforhisexperimentalwork,
valuablesuggestions,andhelpinpreparingthefigures.IalsothanktheHungarian
NationalScienceFoundation(GrantNo.:OTKAK78102)forfinancialsupport.
REFERENCES
1. J. Wang, L. Urban. The impact of early ADME profiling on drug discovery and
developmentstrategy.DrugDiscoveryWorldFall(2004)73‐86.
2. C. Hansch, P.P. Maloney, T. Fujita, R. Muir. Correlation of biological activity of
phenoxyacetic acids with Hammett substituent constants and partition
coefficients.Nature194(1962)178‐180.
3. G.K.Dixon,J.P.Major,M.J.Rice.,HighThroughputScreening:TheNextGeneration
Bios,Oxford,2000.
4. K. Takács‐Novák, G. Völgyi. Physicochemical profiling in drug research. Magyar
KémiaiFolyóirat111(2005)169‐176(Hungrian).
5. B.Testa,H.vandeWaterbeemd,G.Folkers,R.Guy.,PharmacokineticOptimization
inDrugResearch.Biological,Physicochemical,andComputationalStrategies,Wiley‐
VHCA,Zürich,2001.
52
Physicochemicalprofilingindrugresearchanddevelopment
6. L.Di,E.H.Kerns,G.T.Carter.Drug‐likepropertyconceptsinpharmaceuticaldesign.
CurrentPharmaceuticalDesign15(2009)2184‐2194.
7. E.H.Kerns.Highthroughputphysicochemicalprofilingfordrugdiscovery.Journal
ofPharmaceuticalSciences90(2001)1838‐1858.
8. L.Di,E.H.Kerns.Applicationofpharmaceuticalprofilingassaysforoptimizationof
drug‐likeproperties.CurrentOpinioninDrugDiscoveryandDevelopment8(2005)
495‐504.
9. E.H.Kerns,L.Di.Automationinpharmaceuticalprofiling.JournaloftheAssociation
forLaboratoryAutomation10(2005)114‐123.
10. P.J.Taylor.Hydrophobicpropertiesofdrugs.InC.Hansch,P.G.Sammes,J.B.Taylor,
H.Lennernäs.(Eds.),ComprehensiveMedicinalChemistryVol.4,Pergamon,Oxford,
1990,pp241‐294.
11. A. Avdeef. Absorption and Drug Development: Solubility, Permeability and Charge
State.Wiley,NewYork,2003.
12. A. Avdeef, B. Testa. Physicochemical profiling in drug research: a brief survey of
thestate‐of‐the‐artofexperimentaltechniques.CellularandMolecularLifeSciences
59(2002)1681‐1689.
13. E.H. Kerns, L. Di. Physicochemical profiling: overview of the screens. Drug
DiscoveryToday:Technologies1(2004)343‐348.
14. H. Wan, A.G. Holmén. High throughput screening of physicochemical properties
and in vitro ADME profiling in drug discovery. Combinatorial Chemistry and High
ThroughputScreening12(2009)315‐329.
15. J. Wang, S. Skolnik. Recent advances in physicochemical and ADMET profiling in
drugdiscovery.ChemistryandBiodiversity6(2009)1887‐1899.
16. Y. Henchoz, B. Bard, D. Guillarme, P‐A. Carrupt, J‐L. Veuthey, S. Martel. Analytical
tools for the physicochemical profiling of drug candidates to predict
absorption/distribution. Analytical and Bioanalytical Chemistry 394 (2009) 707‐
729.
17. A.Pagliara,M.Reist,S.Geinoz,P‐A.Carrupt,B.Testa.Evaluationandpredictionof
drugpermeation.JournalofPharmacyandPharmacology51(1999)1339‐1357.
18. E. Kerns, L. Li. Drug‐like properties: concepts, structure design and methods: from
ADMEtotoxicityoptimization.Elsevier,SanDiego,2008.
19. J.K. Seydel, M. Wiese. Drug‐membrane interactions. Analysis, drug distribution,
modeling.In:R.Mannhold,H.Kubinyi,G.Folkers.(Eds),MethodsandPrinciplesin
MedicinalChemistry.Wiley‐VCH,Weinheim,2002.
20. P.A. Shore, B.B. Brodie, C.A.M. Hogben. The gastric secretion of drugs: a pH
partitionhypothesis.JournalofPharmacologyandExperimentalTherapeutics119
(1957)361‐369.
21. R.P. Austin, A.M. Davis, C.N. Manners. Partitioning of ionizing molecules between
aqueous buffers and phospholipid vesicles.JournalofPharmaceutical Sciences 84
(1995)1180‐1183.
22. A. Avdeef, K. Box, J.E. Comer, C. Hibbert, K.Y. Tam. pH‐metric logP. 10.
Determinationofvesiclemembrane‐waterpartitioncoefficientsofionizabledrugs.
PharmaceuticalResearch15(1998)209‐215.
53
Chapter1
23. S.D.Krämer,D.Lombardi,A.Primorac,A.V.Thomae,H.Wunderli‐Allenspach.Lipid
bilayer permeation of drug‐like compounds. Chemistry and Biodiversity 6 (2009)
1900‐1916.
24. Y. Sai, A. Tsuji. Transporter‐mediated drug delivery: recent progress and
experimentalapproaches.DrugDiscoveryToday9(2004)712‐720.
25. R. Didziapertis, P. Japertas, A. Avdeef, A. Petrauskas. Classification analysis of P‐
glycoproteinsubstratespecificity.JournalofDrugTargeting11(2003)391‐406.
26. A.Albert,E.P.Serjeant.TheDeterminationofIonizationConstants.3rded.Chapman
&Hall,London,1984.
27. B. Noszál. Acid‐base properties of bioligands. In K. Burger, (Ed): Biocoordination
EquilibriainBiologicallyActiveSystems.EllisHorwood,Chichester,1990,pp18‐55.
28. K. Takács‐Novák, B. Noszál, I. Hermecz, G. Keresztúri, B. Podányi, G. Szász.
Protonation equilibria of quinolone antibacterials. Journal of Pharmaceutical
Sciences79(1990)1023‐1028.
29. K.Mazák,S.Hosztafi,Á.Rácz,B.Noszál.Structuralandphysicochemicalprofilingof
morphine and related compounds of therapeutic interest. Mini‐Reviews in
MedicinalChemistry9(2009)984‐995.
30. K. Kóczián, G. Völgyi, J. Kökösi, B. Noszál. Site‐specific acid‐base properties of
tenoxicam.HelveticaChimicaActa90(2007)1681‐1690.
31. A. Dogan, E. Kilic. Tautomeric and microscopic protonation equilibria of some α‐
aminoacids.AnalyticalBiochemistry365(2007)7‐13.
32. K. Takács‐Novák, B. Noszál, M. Tőkés‐Kövesdi, G. Szász. Acid‐base properties of
terbutaline in terms of protonation macro‐ and microconstants. Journal of
PharmacyandPharmacology47(1995)431‐435.
33. A. Marosi, Z. Kovács, S. Béni, J. Kökösi, B. Noszál. Triprotic acid‐base
microequilibria and pharmacokinetic sequelae of cetirizine. European Journal of
PharmaceuticalSciences37(2009)321‐328.
34. Z. Szakács, M. Kraszni, B. Noszál. Determination of microscopic acid‐base
parameters from NMR‐pH titrations. Analytical and Bioanalytical Chemistry 378
(2004)1428‐1448.
35. N.Sum,A.Avdeef.BiorelevantpKa(37°C)predictedfromthe2Dstructureofthe
moleculeanditspKaat25°C.JournalofPharmaceuticalandBiomedicalAnalysis56
(2011)173‐182.
36. H.H.Cocolas.Antiarrhythmicdrugs.InJ.N.Delgado,W.A.Remers.(Eds),Wilsonand
Gisvold’sTextbookofOrganicMedicinalandPharmaceuticalChemistry.9thedition,
Lippincott,NewYork,1991pp.546‐555.
37. S.H.Yalkowsky,S.Banerjee. AqueousSolubility:MethodsofEstimationforOrganic
Compounds.MarcelDekker,NewYork,1992.
38. D.J.W.Gant,T.Higuchi.SolubilityBehaviorofOrganicCompounds.Wiley,NewYork,
1990.
39. A. Avdeef. Solubility of sparingly‐soluble ionizable drugs. Advanced Drug Delivery
Reviews59(2007)568‐590.
40. C.A.S. Bergström, K. Luthman, P. Artursson. Accuracy of calculated pH‐dependent
aqueous drug solubility. European Journal of Pharmaceutical Sciences 22 (2004)
387‐398.
54
Physicochemicalprofilingindrugresearchanddevelopment
41. A. Avdeef, D. Voloboy, A. Foreman. Dissolution‐solubility: pH, buffer, salt, dual‐
solid, and aggregation effects. In B.Testa, H. van de Waterbeemd. (Eds):
Comprehensive Medicinal Chemistry II. Vol. 5. ADME‐TOX Approaches, Elsevier,
Oxford,2007pp.399‐423.
42. G. Völgyi, E. Baka, K. Box, J.E. Comer, K. Takács‐Novák. Study of pH‐dependent
solubility of organic bases. Revisit of Henderson‐Hasselbalch relationship.
AnalyticaChimicaActa673(2010)40‐46.
43. G.L. Amidon, H. Lennernäs, V.P. Shah, J.R. Crison. A theoretical bases for a
biopharmaceutic drug classification: The correlation of in vitro drug product
dissolution and in vivo bioavailability. Pharmaceutical Research 12 (1995) 413‐
420.
44. J.B.Dressmann,M.Vertzoni,K.Goumas,C.Reppas.Estimatingdrugsolubilityinthe
gastrointestinaltract.AdvancedDrugDeliveryReviews59(2007)591‐602.
45. A. Leo, C. Hansch, D. Elkins. The partition coefficients and their uses. Chemical
Reviews71(1971)525‐616.
46. A.Avdeef.Assessmentofdistribution‐pHprofiles.InV.Pliska,B.Testa,H.vande
Waterbeemd. (Eds), Methods and Principles in Medicinal Chemistry. Vol. 4. CVH,
Weinheim,1996pp.109‐139.
47. K. Takács‐Novák, M. Józan, G. Szász. Lipophilicity of amphoteric molecules
expressedbythetruepartitioncoefficient.InternationalJournalofPharmaceutics
113(1995)47‐55.
48. N.P. Franks,M.H. Abraham, W.R. Lieb. Molecularorganizationof liquid n‐octanol:
An X‐ray diffraction analysis. Journal of Pharmaceutical Sciences 82 (1993) 466‐
470.
49. D.E. Leahy, P.J. Taylor, A.R. Wait. Model solvent systems for QSAR. 1. Propylene
glycol dipelargonate (PGDP). A new standard for use in partition coefficient
determination.QuantitativeStructure‐ActivityRelationships8(1989)17‐31.
50. G.P.van Balen, C.A.M. Martinet, G. Caron, G. Bouchard, M. Reist, P‐A. Carrupt, R.
Fruttero, A. Gasco, B. Testa. Liposome/water lipophilicity: Methods, information
content,andpharmaceuticalapplication.MedicalResearchReviews24(2004)299‐
324.
51. S. Krämer. Liposome/water partitioning: theory and applications. In B. Testa, H.
van de Waterbeemd, G. Folkers, R. Guy. (Eds), Pharmacokinetic Optimization in
Drug Research. Biological, Physicochemical, and Computational Strategies, Wiley‐
VHCA,Zürich,2001.
52. C.A. Lipinski, F. Lombardo B.W. Dominy, P.J. Feeney. Experimental and
computational approaches to estimate solubility and permeability in drug
discoveryanddevelopmentsettings.AdvancedDrugDeliveryReviews23(1997)3‐
25.
53. C. Hansch, J.P. Björkroth, A. Leo. Hydrophobicity and central nervous system
agents: On the principle of minimal hydrophobicity in drug design. Journal of
PharmaceuticalSciences76(1987)663‐687.
54. G. Ferenczy, G.M. Keserű. Thermodynamics guided lead discovery and
optimization.DrugDiscoveryToday15(2010)919‐932.
55. Z.Szakács,G.Hägele.AccuratedeterminationoflowpKvaluesby1HNMRtitration.
Talanta,62(2004)819‐825.
55
Chapter1
56. H. Hegedűs, A. Gergely, P. Horváth, B. Noszál. Acid‐base properties of biological
phenyl‐alkyl‐amines, characterized by CD‐pH titration. Journal of Chemical
ResearchSynopses(1999)306‐307.
57. F.Z. Oumada, C. Ràfols, M. Rosés, E. Bosch. Chromatographic determination of
aqueous dissociation constants of some water‐insoluble nonsteroidal anti‐
inflammatorydrugs.JournalofPharmaceutialSciences91(2002)991‐999.
58. K. Box, J.E. Comer. Using measured pKa, logP and solubility to investigate
supersaturationandpredictBCSclass.CurrentDrugMetabolism9(2008)869‐878.
59. A.Avdeef,J.J.Bucher.Accuratemeasurementoftheconcentrationofhydrogenions
withaglasselectrode:CalibrationsusingthePrideauxandotheruniversalbuffer
solutions and a computer‐controlled automatic titrator. Analytical Chemistry 50
(1978)2137‐2142.
60. J.E.Comer. High‐throughput measurement of logD and pKa. In P. Artursson, H.
Lennernäs, H. van de Warterbeemd. (Eds), Methods and Principles in Medicinal
Chemistry.Vol.18.Wiley‐CVH,Weinheim,2003pp.21‐45.
61. R.I. Allen, K. Box, J.E. Comer, C. Peake, K.Y. Tam. Multiwavelength
spectrophotometric determination of acid dissociation constants of ionisable
drugs.JournalofPharmaceuticalandBiomedicalAnalysis17(1998)699‐712.
62. K.Y.Tam,K.Takács‐Novák.Multiwavelengthspectrophotometricdeterminationof
aciddissociationconstants:avalidationstudy.AnalyticaChimicaActa434(2001)
157‐167.
63. Z. Mandić, V. Gabelica. Ionization, lipophilicity and solubility properties of
repaglinide.JournalofPharmaceuticalandBiomedicalAnalysis41(2006)866‐871.
64. M‐H. Langlois, M. Montagut, J‐P. Dubost, J. Grellet, M‐C. Saux. Protonation
equilibrium and lipophilicity of moxifloxacin. Journal of Pharmaceutical and
BiomedicalAnalysis37(2005)389‐393.
65. M. Tollinger, J.D. Forman‐Kay, L.E. Kay. Measurement of side‐chain carboxyl pKa
valuesofglutamateandaspartateresiduesinanunfoldedproteinbymultinuclear
NMRspectroscopy.JournalofAmericanChemicalSociety124(2002)5714‐5717.
66. Z. Szakács, S. Béni, Z. Varga, L. Őrfi, G. Kéri, B. Noszál. Acid‐base profiling of
imatinibanditsfragments.JournalofMedicinalChemistry48(2005)249‐255.
67. Z. Jia. Physicochemical profiling by capillary electrophoresis. Current
PharmaceuticalAnalysis1(2005)41‐56.
68. S.Babić,A.J.M.Horvat,D.MutavdžićPavlović,M.Kaštelan‐Macan.Determinationof
pKavaluesofactivepharmaceuticalingredients.TrendsinAnalyticalChemistry26
(2007)1043‐1061.
69. S.K. Poole, S. Patel, K. Dehring, H. Workman, C.F. Poole. Determination of acid
dissociation constants by capillary electrophoresis. Journal of Chromatography A
1037(2004)445‐454.
70. S. Pang, J. Kenseth, S. Coldiron. High throughput multiplexed capillary
electrophoresisindrugdiscovery.DrugDiscoveryToday9(2004)1072‐1080.
71. K. Box, J.E. Comer, P. Hosking, K.Y. Tam, L. Trowbridge, A. Hill. Rapid
physicochemicalprofilingasanaidtodrugcandidateselection.InG.K.Dixon,J.S.
Major, M.J. Rice (Eds), High Throughput Screening: The Next Generation. Bios,
Oxford,2000,pp.67‐74.
56
Physicochemicalprofilingindrugresearchanddevelopment
72. K. Box, C. Bevan, J.E. Comer, A. Hill, R. Allen, D. Reynolds. High‐throughput
measurementofpKavaluesinamixed‐bufferlinearpHgradientsystem.Analytical
Chemistry75(2003)883‐892.
73. A.Avdeef,J.E.Comer,S.J.Thomson.pH‐metriclogP.3.Glasselectrodecalibrationin
methanol/water, applied to pKa determination of water‐insoluble substances.
AnalyticalChemistry65(1993)42‐49.
74. K. Takács‐Novák, K.Box, A. Avdeef. Potentiometric pKa determination of water‐
insolublecompounds:validationstudyinmethanol/watermixtures.International
JournalofPharmaceutics151(1997)235‐248.
75. K.Box,G.Völgyi,R.Ruiz,J.E.Comer,K.Takács‐Novák,E.Bosch,C.Ràfols,M.Rosés.
PhysicochemicalpropertiesofanewmulticomponentcosolventsystemforthepKa
determination of poorly soluble pharmaceutical compounds. Helvetica Chimica
Acta90(2007)1538‐1553.
76. G.Völgyi,R.Ruiz,K.Box,J.E.Comer,E.Bosch,K.Takács‐Novák.Potentiometricand
spectrophotometric pKa determination of water‐insoluble compounds: Validation
studyinanewcosolventsystem.AnalyticaChimicaActa583(2007)418‐428.
77. C.Lipinski.Computationalandexperimentalapproachestoavoidingsolubilityand
oralabsorptionproblemsinearlydiscovery.InR.Borchard(Ed),DesigningDrugs
withOptimalinVivoActivityafterOraladministration.DrewUniversityResidential
SchoolonMedicinalChemistry,Madison,NJ.2000.
78. C.Bevan,R.S.Lloyd.Ahighthroughputscreeningmethodforthedeterminationof
aqueoussolubilitybylasernephelometryinmicrotitreplates.AnalyticalChemistry
72(2000)1781‐1787.
79. A.Avdeef.High‐throughputmeasurementsofsolubilityprofiles.InB.Testa,H.van
de Waterbeemd, G. Folkers, R. Guy (Eds), Pharmacokinetic Optimization in Drug
Research. Biological, Physicochemical, and Computational Strategies, Wiley‐VHCA,
Zürich,2001.pp.305‐326.
80. T. Yamashita, Y. Dohta, T. Nakamura, T. Fukami. High‐speed solubility screening
assayusingultra‐performanceliquidchromatography/massspectrometryindrug
discovery.JournalofChromatographyA1182(2008)72‐76.
81. E.Baka,J.E.Comer,K.Takács‐Novák.Studyofequilibriumsolubilitymeasuredby
saturation shake‐flask method using hydrochlorothiazide as model compound.
JournalofPharmaceuticalandBiomedicalAnalysis46(2008)335‐341.
82. E. Baka. Good laboratory practice of equilibrium solubility measurements. Acta
PharmaceuticaHungarica81(2011)18‐28(Hungarian).
83. A.Avdeef.pH‐metricsolubility.1.SolubilitypHprofilesfromBjerrumplots.Gibbs
buffer and pKa in the solid state. Pharmacy and Pharmacology Communications 4
(1998)165‐178.
84. A. Avdeef, C.M. Berger, C. Brownell. pH‐metric solubility. 2. Correlation between
the acid‐base titration and the saturation shake‐flask solubility‐pH methods.
PharmaceuticalResearch17(2000)85‐89.
85. A. Avdeef, C.M. Berger. pH‐metric solubility. 3. Dissolution titration template
methodforsolubilitydetermination.EuropeanJournalofPharmaceuticalSciences
14(2001)281‐291.
86. M.Stuart,K.Box.Chasingequilibrium:measuringoftheintrinsicsolubilityofweak
acidsandbases.AnalyticalChemistry77(2005)983‐990.
57
Chapter1
87. K. Box, G. Völgyi, E. Baka, M. Stuart, K. Takács‐Novák, J.E. Comer. Equilibrium
versuskineticmeasurementsofaqueoussolubilityandtheabilityofcompoundsto
supersaturateinsolution.Avalidationstudy.JournalofPharmaceuticalSciences95
(2006)1298‐1307.
88. A. Avdeef, O. Tsinman. Miniaturized rotating disk intrinsic dissolution rate
measurement: Effect of buffer capacity in comparisons to traditional Wood’s
apparatus.PharmaceuticalResearch25(2008)2613‐2627.
89. K. Tsinman, A. Avdeef , O. Tsinman, D. Voloboy. Powder dissolution method for
estimating rotating disk intrinsic dissolution rates of low solubility drugs.
PharmaceuticalResearch26(2009)2093‐2100.
90. A. Glomme, J. März, J.B. Dressman. Comparison of a miniaturized shake‐flask
solubility method with automated potentiometric acid/base titrations and
calculatedsolubilities.JournalofPharmaceuticalSciences94(2005)1‐16.
91. A. Glomme, J. März, J.B. Dressman. Predicting the intestinal solubility of poorly
soluble molecules. In B. Testa, S.D. Krämer, H. Wunderli‐Allenspach, G. Folkers
(Eds),PharmacokineticProfilinginDrugResearch:Biological,Physicochemicaland
ComputationalStrategies,Wiley‐VCH,Weinheim,2006.pp259‐280.
92. L. Zhou, L. Yang, L., S. Tilton, J. Wang. Development of a high throughput
equilibrium solubility assay using miniaturized shake‐flask method in early drug
discovery.JournalofPharmaceuticalSciences96(2007)3052‐3071.
93. A.T.Serajuddin.Saltformationtoimprovedrugsolubility.AdvancedDrugDelivery
Reviews59(2007)603‐616.
94. J. Alsenz, E. Meister, E. Haenel. Development of a partially automated solubility
screening (PASS) assay for early drug development. Journal of Pharmaceutical
Sciences96(2007)1748‐1762.
95. A.Llinàs,K.Box,J.Burley,R.Glen,J.Goodman.Anewmethodforthereproducible
generation of polymorphs: two forms of sulindac with very different solubilities.
JournalofAppliedCrystallography40(2007)379‐381.
96. B.Bard,S.Martel,P‐A.Carrupt.HighthroughputUVmethodfortheestimationof
thermodynamic solubility and the determination of the solubility in biorelevant
media.EuropeanJournalofPharmaceuticalSciences33(2008)230‐240.
97. J.Fagenberg,O.Tsinman,N.Sun,K.Tsinman,A.Avdeef,C.Bergström.Dissolution
rate and apparent solubility of poorly soluble drugs in biorelevant dissolution
media.MolecularPharmaceutics7(2010)1419‐1430.
98. A. Hersey, A. Hill, R. Hyde, D. Livingstone. Principles of method selection in
partitionstudies.QuantitativeStructure‐ActivityRelationships8(1989)288‐296.
99. L‐G. Danielsson, Y‐H. Zhang. Methods for determining n‐octanol‐water partition
constants.TrendsinAnalyticalChemistry15(1996)188‐196.
100. J.E.Comer,K.Tam.Lipophilicityprofiles:theoryandmeasurement.InB.Testa,B.,
H. van de Waterbeemd, G. Folkers, R. Guy (Eds), Pharmacokinetic Optimization in
Drug Research. Biological, Physicochemical, and Computational Strategies, Wiley‐
VHCA,Zürich,2001.pp.275‐304.
101. J.C.Dearden,G.M.Bresnen.Themeasurementofpartitioncoefficient.Quantitative
Structure‐ActivityRelationships7(1988)133‐144.
58
Physicochemicalprofilingindrugresearchanddevelopment
102. K. Takács‐Novák. Practical aspects of determination of partition coefficient
according to GLP rules. Acta Pharmaceutica Hungarica 67 (1997) 179‐191
(Hungarian).
103. B. Slater, A. McCormack, A. Avdeef, J.E. Comer. pH‐metric logP. 4. Comparison of
partition coefficients determined by shake‐flask, HPLC and potentiometric
methods.JournalofPharmaceuticalSciences83(1994)1280‐1283.
104. K.Takács‐Novák,A.Avdeef.InterlaboratorystudyoflogPdeterminationbyshake‐
flask and potentiometric methods. Journal of Pharmaceutical and Biomedical
Analysis14(1996)1405‐1413.
105. K. Valkó. Application of high‐performance liquid chromatography based
measurements of lipophilicity to model biological distribution. Journal of
ChromatographyA1037(2004)299‐310.
106. R. Kaliszan. QSRR: Quantitative structure‐(chromatographic) retention
relationships.ChemicalReviews107(2007)3212‐3246.
107. G.L.Biagi,A.M.Barbaro,A.Sapone,M.Recenatini.Determinationoflipophilicityby
means of reversed‐phase thin‐layer chromatography. I. Basic aspects and
relationship between slope and intercept in TLC equations. Journal of
ChromatographyA662(1994)341‐361.
108. A.Pyka,M.Babuska.Lipophilicityofselectedsteroidcompounds.I.Investigations
on RP18W stationary phase by RP‐HPTLC. Journal of Liquid Chromatography and
RelatedTechnologies29(2006)1891‐1903.
109. Z. Rozmer, P. Perjési, K. Takács‐Novák. Use of RP‐TLC for logP determination of
isomeric chalcones and cyclic chalcone analogues. Journal of Planar
Chromatography19(2006)124‐128.
110. G. Völgyi, K. Deák, J. Vámos, K. Valkó, K. Takács‐Novák. RPTLC determination of
logPofstructurallydiverseneutralcompounds.JournalofPlanarChromatography
21(2008)143‐149.
111. L. Hitzel, A.P. Watt, K.L. Locker. An increased throughput method for the
determinationofpartitioncoefficients.PharmaceuticalResearch17(2000)1389‐
1395.
112. D.M.Wilson,X.Wang,E.Walsh,R.A.Rourick.HighthroughputlogDdetermination
using liquid chromatography‐mass spectrometry. Combinatorial Chemistry and
HighThroughputScreening4(2001)511‐519.
113. K. Takács‐Novák. Investigation of physicochemical properties of deramciclane
(EGIS‐3886), a new anxiolytic compound. Acta Pharmaceutica Hungarica 69
(1999)123‐127(Hungarian).
114. K.Takács‐Novák,P.Perjési,J.Vámos.DeterminationoflogPforbiologicallyactive
chalcones and chalcone analogs by RPTLC. Journal of Planar Chromatography 14
(2001)42‐46.
59